34:null 35:["$","title",null,{"children":"Project 1: Univariate and Bivariate Data | Jose.carmona-sanchez | Observable"}] 3b:Tf34,let all_task_data; { // 2. Correct syntax for Observable Framework npm imports const { default: JSZip } = await import("npm:jszip@3"); // 3. Inlined fallback function to parse each individual NumPy array structure const parseNpy = (buffer) => { const headerLength = new DataView(buffer.slice(8, 10)).getUint16(0, true); const headerBuffer = buffer.slice(10, 10 + headerLength); const headerText = new TextDecoder("utf-8").decode(headerBuffer); const descrMatch = headerText.match(/'descr':\s*'([^']+)'/); const shapeMatch = headerText.match(/'shape':\s*$([^)]+)$/); const dtype = descrMatch ? descrMatch[1] : "|u1"; const shape = shapeMatch ? shapeMatch[1].split(",").map(Number).filter(Boolean) : []; const dataBuffer = buffer.slice(10 + headerLength); let data; if (dtype.includes("f4")) data = new Float32Array(dataBuffer); else if (dtype.includes("f8")) data = new Float64Array(dataBuffer); else if (dtype.includes("i4")) data = new Int32Array(dataBuffer); else data = new Uint8Array(dataBuffer); return { dtype, shape, data }; }; // 4. Read the local file deployment payload const npzBuffer = await FileAttachment("all-task_data.npz").arrayBuffer(); const zip = await JSZip.loadAsync(npzBuffer); // 5. Build your temporary object locally const parsedData = {}; // 6. Iterate over every array file packed inside the .npz archive for (const [filename, file] of Object.entries(zip.files)) { if (filename.endsWith('.npy')) { const arrayBuffer = await file.async("arraybuffer"); const key = filename.replace('.npy', ''); parsedData[key] = parseNpy(arrayBuffer); } } // 7. Assign the finished data to the top-level variable and display it all_task_data = parsedData; display(all_task_data); } let all_task_data_lobes; { // 2. Correct syntax for Observable Framework npm imports const { default: JSZip } = await import("npm:jszip@3"); // 3. Inlined fallback function to parse each individual NumPy array structure const parseNpy = (buffer) => { const headerLength = new DataView(buffer.slice(8, 10)).getUint16(0, true); const headerBuffer = buffer.slice(10, 10 + headerLength); const headerText = new TextDecoder("utf-8").decode(headerBuffer); const descrMatch = headerText.match(/'descr':\s*'([^']+)'/); const shapeMatch = headerText.match(/'shape':\s*$([^)]+)$/); const dtype = descrMatch ? descrMatch[1] : "|u1"; const shape = shapeMatch ? shapeMatch[1].split(",").map(Number).filter(Boolean) : []; const dataBuffer = buffer.slice(10 + headerLength); let data; if (dtype.includes("f4")) data = new Float32Array(dataBuffer); else if (dtype.includes("f8")) data = new Float64Array(dataBuffer); else if (dtype.includes("i4")) data = new Int32Array(dataBuffer); else data = new Uint8Array(dataBuffer); return { dtype, shape, data }; }; // 4. Read the local file deployment payload const npzBuffer = await FileAttachment("all-task_data_lobes.npz").arrayBuffer(); const zip = await JSZip.loadAsync(npzBuffer); // 5. Build your temporary object locally const parsedData = {}; // 6. Iterate over every array file packed inside the .npz archive for (const [filename, file] of Object.entries(zip.files)) { if (filename.endsWith('.npy')) { const arrayBuffer = await file.async("arraybuffer"); const key = filename.replace('.npy', ''); parsedData[key] = parseNpy(arrayBuffer); } } // 7. Assign the finished data to the top-level variable and display it all_task_data_lobes = parsedData; display(all_task_data_lobes); } const all_task_within_lobes = await FileAttachment("all_within_lobe.json").json(); display(all_task_within_lobes); const all_task_across_lobes = await FileAttachment("all_across_lobe.json").json(); display(all_task_across_lobes);3c:Tf2e,let task_film_data; { // 2. Correct syntax for Observable Framework npm imports const { default: JSZip } = await import("npm:jszip@3"); // 3. Inlined fallback function to parse each individual NumPy array structure const parseNpy = (buffer) => { const headerLength = new DataView(buffer.slice(8, 10)).getUint16(0, true); const headerBuffer = buffer.slice(10, 10 + headerLength); const headerText = new TextDecoder("utf-8").decode(headerBuffer); const descrMatch = headerText.match(/'descr':\s*'([^']+)'/); const shapeMatch = headerText.match(/'shape':\s*$([^)]+)$/); const dtype = descrMatch ? descrMatch[1] : "|u1"; const shape = shapeMatch ? shapeMatch[1].split(",").map(Number).filter(Boolean) : []; const dataBuffer = buffer.slice(10 + headerLength); let data; if (dtype.includes("f4")) data = new Float32Array(dataBuffer); else if (dtype.includes("f8")) data = new Float64Array(dataBuffer); else if (dtype.includes("i4")) data = new Int32Array(dataBuffer); else data = new Uint8Array(dataBuffer); return { dtype, shape, data }; }; // 4. Read the local file deployment payload const npzBuffer = await FileAttachment("task-film_data.npz").arrayBuffer(); const zip = await JSZip.loadAsync(npzBuffer); // 5. Build your temporary object locally const parsedData = {}; // 6. Iterate over every array file packed inside the .npz archive for (const [filename, file] of Object.entries(zip.files)) { if (filename.endsWith('.npy')) { const arrayBuffer = await file.async("arraybuffer"); const key = filename.replace('.npy', ''); parsedData[key] = parseNpy(arrayBuffer); } } // 7. Assign the finished data to the top-level variable and display it task_film_data = parsedData; display(task_film_data); } let task_film_data_lobes; { // 2. Correct syntax for Observable Framework npm imports const { default: JSZip } = await import("npm:jszip@3"); // 3. Inlined fallback function to parse each individual NumPy array structure const parseNpy = (buffer) => { const headerLength = new DataView(buffer.slice(8, 10)).getUint16(0, true); const headerBuffer = buffer.slice(10, 10 + headerLength); const headerText = new TextDecoder("utf-8").decode(headerBuffer); const descrMatch = headerText.match(/'descr':\s*'([^']+)'/); const shapeMatch = headerText.match(/'shape':\s*$([^)]+)$/); const dtype = descrMatch ? descrMatch[1] : "|u1"; const shape = shapeMatch ? shapeMatch[1].split(",").map(Number).filter(Boolean) : []; const dataBuffer = buffer.slice(10 + headerLength); let data; if (dtype.includes("f4")) data = new Float32Array(dataBuffer); else if (dtype.includes("f8")) data = new Float64Array(dataBuffer); else if (dtype.includes("i4")) data = new Int32Array(dataBuffer); else data = new Uint8Array(dataBuffer); return { dtype, shape, data }; }; // 4. Read the local file deployment payload const npzBuffer = await FileAttachment("task-film_data_lobes.npz").arrayBuffer(); const zip = await JSZip.loadAsync(npzBuffer); // 5. Build your temporary object locally const parsedData = {}; // 6. Iterate over every array file packed inside the .npz archive for (const [filename, file] of Object.entries(zip.files)) { if (filename.endsWith('.npy')) { const arrayBuffer = await file.async("arraybuffer"); const key = filename.replace('.npy', ''); parsedData[key] = parseNpy(arrayBuffer); } } // 7. Assign the finished data to the top-level variable and display it task_film_data_lobes = parsedData; display(task_film_data_lobes); } const film_within_lobes = await FileAttachment("film_within_lobe.json").json(); display(film_within_lobes); const film_across_lobes = await FileAttachment("film_across_lobe.json").json(); display(film_across_lobes);3d:Tf2e,let task_rest_data; { // 2. Correct syntax for Observable Framework npm imports const { default: JSZip } = await import("npm:jszip@3"); // 3. Inlined fallback function to parse each individual NumPy array structure const parseNpy = (buffer) => { const headerLength = new DataView(buffer.slice(8, 10)).getUint16(0, true); const headerBuffer = buffer.slice(10, 10 + headerLength); const headerText = new TextDecoder("utf-8").decode(headerBuffer); const descrMatch = headerText.match(/'descr':\s*'([^']+)'/); const shapeMatch = headerText.match(/'shape':\s*$([^)]+)$/); const dtype = descrMatch ? descrMatch[1] : "|u1"; const shape = shapeMatch ? shapeMatch[1].split(",").map(Number).filter(Boolean) : []; const dataBuffer = buffer.slice(10 + headerLength); let data; if (dtype.includes("f4")) data = new Float32Array(dataBuffer); else if (dtype.includes("f8")) data = new Float64Array(dataBuffer); else if (dtype.includes("i4")) data = new Int32Array(dataBuffer); else data = new Uint8Array(dataBuffer); return { dtype, shape, data }; }; // 4. Read the local file deployment payload const npzBuffer = await FileAttachment("task-rest_data.npz").arrayBuffer(); const zip = await JSZip.loadAsync(npzBuffer); // 5. Build your temporary object locally const parsedData = {}; // 6. Iterate over every array file packed inside the .npz archive for (const [filename, file] of Object.entries(zip.files)) { if (filename.endsWith('.npy')) { const arrayBuffer = await file.async("arraybuffer"); const key = filename.replace('.npy', ''); parsedData[key] = parseNpy(arrayBuffer); } } // 7. Assign the finished data to the top-level variable and display it task_rest_data = parsedData; display(task_rest_data); } let task_rest_data_lobes; { // 2. Correct syntax for Observable Framework npm imports const { default: JSZip } = await import("npm:jszip@3"); // 3. Inlined fallback function to parse each individual NumPy array structure const parseNpy = (buffer) => { const headerLength = new DataView(buffer.slice(8, 10)).getUint16(0, true); const headerBuffer = buffer.slice(10, 10 + headerLength); const headerText = new TextDecoder("utf-8").decode(headerBuffer); const descrMatch = headerText.match(/'descr':\s*'([^']+)'/); const shapeMatch = headerText.match(/'shape':\s*$([^)]+)$/); const dtype = descrMatch ? descrMatch[1] : "|u1"; const shape = shapeMatch ? shapeMatch[1].split(",").map(Number).filter(Boolean) : []; const dataBuffer = buffer.slice(10 + headerLength); let data; if (dtype.includes("f4")) data = new Float32Array(dataBuffer); else if (dtype.includes("f8")) data = new Float64Array(dataBuffer); else if (dtype.includes("i4")) data = new Int32Array(dataBuffer); else data = new Uint8Array(dataBuffer); return { dtype, shape, data }; }; // 4. Read the local file deployment payload const npzBuffer = await FileAttachment("task-rest_data_lobes.npz").arrayBuffer(); const zip = await JSZip.loadAsync(npzBuffer); // 5. Build your temporary object locally const parsedData = {}; // 6. Iterate over every array file packed inside the .npz archive for (const [filename, file] of Object.entries(zip.files)) { if (filename.endsWith('.npy')) { const arrayBuffer = await file.async("arraybuffer"); const key = filename.replace('.npy', ''); parsedData[key] = parseNpy(arrayBuffer); } } // 7. Assign the finished data to the top-level variable and display it task_rest_data_lobes = parsedData; display(task_rest_data_lobes); } const rest_within_lobes = await FileAttachment("rest_within_lobe.json").json(); display(rest_within_lobes); const rest_across_lobes = await FileAttachment("rest_across_lobe.json").json(); display(rest_across_lobes);3e:T5fa,### Introduction & Intuition There is an inherent compromise between high temporal and high spatial resolution in current human brain recordings. One of the most prominent neuroimaging techniques is electrophysiology which records electrical activity within the brain (eg EEG, ECoG, iEEG). Electrocorticography (ECoG) is characterized by both high temporal and spatial resolution although due to the invasive nature of ECoG it has minimal brain coverage for an individual. The Gaussian Process Regression (GPR) model shown in this project aims to create a technique that has high temporal and spatial resolution while possessing a relatively large brain coverage. ![Intuition behind the GPR model, showing how correlation matrix can be used to infer electrical brain activity](${await FileAttachment('dropped.png').url()}) The image shown above gives a good intuition to how the GPR model functions. If we know correlation between locations A and B (patient 1) and we know correlation between location B and C (patient 2) than we can leverage that information to infer activity at missing location. In this project the idea of inferring brain activity is also called reconstruction. In the sense that for a given electrode location we are reconstructing the electrical value (mV) that an electrode would've empirically picked up. Here I test the hypothesis: **Given more data, regardless of task type the GPR model should have a larger quantity of highly correlated reconstructed correlation coefficients.**3f:Td7c,// Put all correlation data together in form ready to pull const allData = [ ...Array.from(all_task_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "All Task" })), ...Array.from(all_task_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "All Task" })), ...Array.from(task_rest_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "Rest Task" })), ...Array.from(task_rest_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "Rest Task" })), ...Array.from(task_film_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "Film Task" })), ...Array.from(task_film_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "Film Task" })) ]; // set correlation domain const corrDomain = [-1,1]; //set colors for within vs acorss const groupColor = { domain: ["within", "across"], range: ["#c24806", "#5406c2"] }; //plot histogram — faceting on fy: "group" gives within/across their own row //so the bars are never drawn on top of each other const histPlot = Plot.plot({ width: 1000, height: 550, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { grid: true, label: "Frequency", nice: true }, color: { ...groupColor, legend: true }, fx: { domain: ["All Task", "Rest Task", "Film Task"], label: null }, fy: { domain: ["within", "across"], label: null }, marks: [ Plot.rectY( allData, Plot.binX({ y: "count" }, { x: "value", fill: "group", fx: "dataset", fy: "group" }) ), Plot.ruleY([0]) ] }); //plot box plots down below histogram const boxPlot = Plot.plot({ width: 1000, height: 120, marginBottom: 35, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { label: null, domain: ["within", "across"] }, color: groupColor, fx: { domain: ["All Task", "Rest Task", "Film Task"], label: null }, marks: [ Plot.boxX(allData, { x: "value", y: "group", stroke: "group", fx: "dataset", strokeWidth: 2, fill: "group", fillOpacity: 0.2 }) ] }); display(html`

GPR Model Reconstruction Correlation Distribution for Different Task Types

${histPlot} ${boxPlot}

The figure shown above displays the distribution of correlation coefficients as a result of reconstruction from three different GPR models that was trained off different sub-datasets that was seperated by task. Within the orginal data there was either resting data (participants had no external audio or visual stimuli) or the film data (patricipants would watch a film with sound). The models were trained off just resting data (rest task), film data (film task), or both (all task). With the three models there is also an acorss and within model. The within model is created with only the indivdual subjects data. While the across model is created by pooling all data together from all subjects. To ensure validity of results of the reconstruction a leave one in leave one out method is used, more about this method can be read here.

`);40:T561,const lobeOrder = [ "Out_of_Bounds", "Frontal", "Temporal", "Parietal", "Occipital", "Cingulate_Cortex", "Subcortical/Cerebellum", "Other/Deep_Structure" ]; // Build paired {region, value} arrays, using matching indices between the // across_corr correlation values and their corresponding lobe labels. function makeLobeData(corrData, lobeLabels) { const corr = Array.from(corrData); return corr.map((v, i) => ({ region: lobeLabels[i], value: v })); } const allTaskLobeData = makeLobeData(all_task_data_lobes.across_corr.data, all_task_across_lobes); const filmLobeData = makeLobeData(task_film_data_lobes.across_corr.data, film_across_lobes); const restLobeData = makeLobeData(task_rest_data_lobes.across_corr.data, rest_across_lobes); display({allTaskLobeData_len: allTaskLobeData.length, filmLobeData_len: filmLobeData.length, restLobeData_len: restLobeData.length}); const within_allTaskLobeData = makeLobeData(all_task_data_lobes.within_corr.data, all_task_within_lobes); const within_filmLobeData = makeLobeData(task_film_data_lobes.within_corr.data, film_within_lobes); const within_restLobeData = makeLobeData(task_rest_data_lobes.within_corr.data, rest_within_lobes); display({within_allTaskLobeData_len: within_allTaskLobeData.length, within_filmLobeData_len: within_filmLobeData.length, within_restLobeData_len: within_restLobeData.length});41:Tf3a, function getStats2(values) { const sorted = values.slice().sort((a, b) => a - b); const n = sorted.length; const q1 = sorted[Math.floor(n * 0.25)]; const q3 = sorted[Math.floor(n * 0.75)]; const iqr = q3 - q1; const lowerFence = q1 - 1.5 * iqr; const upperFence = q3 + 1.5 * iqr; //Determine min and max values/upper and lower limit const withinFences = values.filter(v => v >= lowerFence && v <= upperFence); const min = withinFences.length > 0 ? Math.min(...withinFences) : q1; const max = withinFences.length > 0 ? Math.max(...withinFences) : q3; return { min: min, q1: q1, median: sorted[Math.floor(n * 0.5)], q3: q3, max: max, n: n, lowerFence: lowerFence, upperFence: upperFence }; } function make_Scatter2(data, title, subtitle) { const typeOrder = ["within", "across"]; const statsData = typeOrder.flatMap(type => lobeOrder.map(region => { const values = data.filter(d => d.region === region && d.type === type).map(d => d.value); if (values.length === 0) return null; const stats = getStats2(values); return { region: region, type: type, ...stats }; }).filter(d => d !== null) ); return Plot.plot({ width: 1500, height: 600, marginBottom: 80, x: { domain: lobeOrder, label: null, tickRotate: -30 }, y: { domain: [-1, 1], grid: true, label: "Correlation" }, title, subtitle, marks: [ Plot.ruleY([0], { stroke: "#999" }), Plot.dot(data, { x: "region", y: "value", fx: "type", fill: "region", r: 2, fillOpacity: 0.35, tip: true }), Plot.boxY(data, { x: "region", y: "value", fx: "type", stroke: "region", strokeWidth: 1.5, fill: "region", stroke: "region", fillOpacity: 0.3 }), Plot.ruleY(statsData, { x1: (d) => d.region, x2: (d) => d.region, y: "min", fx: "type", stroke: "region", strokeWidth: 1, strokeOpacity: 1 }), Plot.ruleY(statsData, { x1: (d) => d.region, x2: (d) => d.region, y: "max", fx: "type", stroke: "region", strokeWidth: 1, strokeOpacity: 1 }), // n at bottom Plot.text(statsData, { x: "region", y: -1, fx: "type", text: d => `n=${d.n}`, fill: "#999", fontSize: 10, dy: -15, textAnchor: "middle" }), ] }); } // Code to Plot da plots const combinedAllTaskData = [ ...allTaskLobeData.map(d => ({ ...d, type: "across" })), ...within_allTaskLobeData.map(d => ({ ...d, type: "within" })) ]; const combinedRestData = [ ...restLobeData.map(d => ({ ...d, type: "across" })), ...within_restLobeData.map(d => ({ ...d, type: "within" })) ]; const combinedFilmData = [ ...filmLobeData.map(d => ({ ...d, type: "across" })), ...within_filmLobeData.map(d => ({ ...d, type: "within" })) ]; const allTaskScatter = display(make_Scatter2( combinedAllTaskData, "All Task — Within vs. Across-Model Reconstruction Correlation by Brain Region", `n = ${within_allTaskLobeData.length} (within) / ${allTaskLobeData.length} (across) locations` )); const restScatter = display(make_Scatter2( combinedRestData, "Rest Task — Within vs. Across-Model Reconstruction Correlation by Brain Region", `n = ${within_restLobeData.length} (within) / ${restLobeData.length} (across) locations` )); const filmScatter = display(make_Scatter2( combinedFilmData, "Film Task — Within vs. Across-Model Reconstruction Correlation by Brain Region", `n = ${within_filmLobeData.length} (within) / ${filmLobeData.length} (across) locations` )); 42:T4aa,### Results and Critiques: Brain Region Reconstruction The hypothesis was correct, depending on the task type that alters the distribution of correlation coefficients when comparing rest and film. Although it is rather interesting how the film and all data is heavily biased towards the negative correlation for all brain regions. Some critiques that I have; I would have liked to placed the within and across results on the same graphs rather than have two separate graphs. I also believe that a better visualization would have been to use violin graphs rather than scatter + boxplots although unfortunately I did not have time to figure out how to do it. My biggest critique is the preprocessing of the data to get the brain regions. About half of the correlation coefficients had to be removed due to the mask I was using to classify where in the brain the electrode coordinates was placed in (went from a total of 6000 reconstructions to 3000). Although I did check the distribution of coefficients with the lower number of samples and it was very similar to the original graph (first graph shown). If I had more time I would go back and alter the preprocessing step to conserve more data. 43:T4f2,### AI Use AI was used in this project to import/use data and familiarize myself with the plotting functions. The AI used in this project was a combination of Deepseek and the integrated AI on observable. By far the toughest part about the project was getting the data imported into observable which is where AI took over cause it was taking too long doing it myself. here are the specific prompts I used: - How to import npz files with multiple numpy arrays inside into observable javascript? As stated this was the toughest part of the project for me so the output for this was not altered from the AI. - can you tell me more about the Observable Plot library and Plot.plot. General code for histogram and boxplot was pulled but was modified so that it would be compatible to use with across and within data would - Best way to plot two datasets together with observable plot library? This line was used when making the univariate graph and it gave me the idea of flattening all data into a single array to create AllData array although this took trial and error to figure out the labels which was important for creating the three histogram and boxplots to be together. This information also came in handy with the scatterplots to create a across and within plot.44:T546,// Put all correlation data together in form ready to pull const alltask_within_data = [ ...Array.from(all_task_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "All Task" })), ]; // set correlation domain const corrDomain = [-1,1]; //set colors for within vs acorss const color = {domain: ["within"],range: ["#c24806"]}; //plot histogram const histPlot = Plot.plot({ width: 1000, height: 450, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { grid: true, label: "Frequency", nice: true }, color: { ...color, legend: true }, marks: [ Plot.rectY( alltask_within_data, Plot.binX({ y: "count" }, { x: "value", fill: "group", mixBlendMode: "multiply", }) ), Plot.ruleY([0]) ] }); //plot box plots down below histogram const boxPlot = Plot.plot({ width: 1000, height: 120, marginBottom: 20, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { label: null}, color: color, marks: [ Plot.boxX(alltask_within_data, { x: "value", y: "group", stroke: "group", strokeWidth: 2, fill: "group", fillOpacity: 0.2 }) ] }); display(html`

GPR Model Reconstruction Correlation Distribution for All Task (Within model)

${histPlot} ${boxPlot}

`);45:T557,const alltask_data = [ ...Array.from(all_task_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "All Task" })), ...Array.from(all_task_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "All Task" })) ]; const corrDomain = [-1,1]; const groupColor = { domain: ["within", "across"], range: ["#c24806", "#5406c2"] }; //plot histogram const histPlot = Plot.plot({ width: 1000, height: 450, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { grid: true, label: "Frequency", nice: true }, color: { ...groupColor, legend: true }, marks: [ Plot.rectY( alltask_data, Plot.binX({ y: "count" }, { x: "value", fill: "group", mixBlendMode: "multiply", }) ), Plot.ruleY([0]) ] }); const boxPlot = Plot.plot({ width: 1000, height: 120, marginBottom: 35, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { label: null, domain: ["within", "across"] }, color: groupColor, marks: [ Plot.boxX(alltask_data, { x: "value", y: "group", stroke: "group", strokeWidth: 2, fill: "group", fillOpacity: 0.2 }) ] }); display(html`

GPR Model Reconstruction Correlation Distribution for All Tasks (Within and Across Models)

${histPlot} ${boxPlot}

`);46:T869,// Put all correlation data together in form ready to pull const allData = [ ...Array.from(all_task_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "All Task" })), ...Array.from(all_task_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "All Task" })), ...Array.from(task_rest_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "Rest Task" })), ...Array.from(task_rest_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "Rest Task" })), ...Array.from(task_film_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "Film Task" })), ...Array.from(task_film_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "Film Task" })) ]; // set correlation domain const corrDomain = [-1,1]; //set colors for within vs acorss const groupColor = { domain: ["within", "across"], range: ["#c24806", "#5406c2"] }; //plot histogram const histPlot = Plot.plot({ width: 1000, height: 450, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { grid: true, label: "Frequency", nice: true }, color: { ...groupColor, legend: true }, fx: { domain: ["All Task", "Rest Task", "Film Task"], label: null }, marks: [ Plot.rectY( allData, Plot.binX({ y: "count" }, { x: "value", fill: "group", mixBlendMode: "multiply", fx: "dataset" }) ), Plot.ruleY([0]) ] }); //plot box plots down below histogram const boxPlot = Plot.plot({ width: 1000, height: 120, marginBottom: 35, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { label: null, domain: ["within", "across"] }, color: groupColor, fx: { domain: ["All Task", "Rest Task", "Film Task"], label: null }, marks: [ Plot.boxX(allData, { x: "value", y: "group", stroke: "group", fx: "dataset", strokeWidth: 2, fill: "group", fillOpacity: 0.2 }) ] }); display(html`

GPR Model Reconstruction Correlation Distribution for Different Task Types

${histPlot} ${boxPlot}

`);47:T6f8,const rest_and_film_data = [ ...Array.from(task_rest_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "Rest Task" })), ...Array.from(task_rest_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "Rest Task" })), ...Array.from(task_film_data.within_corr.data).map(v => ({ value: v, group: "within", dataset: "Film Task" })), ...Array.from(task_film_data.across_corr.data).map(v => ({ value: v, group: "across", dataset: "Film Task" })) ]; const corrDomain = [-1,1]; const groupColor = { domain: ["within", "across"], range: ["#c24806", "#5406c2"] }; const histPlot = Plot.plot({ width: 1000, height: 450, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { grid: true, label: "Frequency", nice: true }, color: { ...groupColor, legend: true }, fx: { domain: ["Rest Task", "Film Task"], label: null }, marks: [ Plot.rectY( rest_and_film_data, Plot.binX({ y: "count" }, { x: "value", fill: "group", mixBlendMode: "multiply", fx: "dataset", fy: "group" }) ), Plot.ruleY([0]) ] }); const boxPlot = Plot.plot({ width: 1000, height: 120, marginBottom: 35, x: { domain: corrDomain, grid: true, label: "Correlation" }, y: { label: null, domain: ["within", "across"] }, color: groupColor, fx: { domain: ["Rest Task", "Film Task"], label: null }, marks: [ Plot.boxX(rest_and_film_data, { x: "value", y: "group", stroke: "group", fx: "dataset", strokeWidth: 2, fill: "group", fillOpacity: 0.2 }) ] }); display(html`

GPR Model Reconstruction Correlation Distribution for Different Task Type

${histPlot} ${boxPlot}

`);48:T565,/*const lobeOrder = [ "Out_of_Bounds", "Frontal", "Temporal", "Parietal", "Occipital", "Cingulate_Cortex", "Subcortical/Cerebellum", "Other/Deep_Structure" ]; // Build paired {region, value} arrays, using matching indices between the // across_corr correlation values and their corresponding lobe labels. function makeLobeData(corrData, lobeLabels) { const corr = Array.from(corrData); return corr.map((v, i) => ({ region: lobeLabels[i], value: v })); } const allTaskLobeData = makeLobeData(all_task_data_lobes.across_corr.data, all_task_across_lobes); const filmLobeData = makeLobeData(task_film_data_lobes.across_corr.data, film_across_lobes); const restLobeData = makeLobeData(task_rest_data_lobes.across_corr.data, rest_across_lobes); display({allTaskLobeData_len: allTaskLobeData.length, filmLobeData_len: filmLobeData.length, restLobeData_len: restLobeData.length}); const within_allTaskLobeData = makeLobeData(all_task_data_lobes.within_corr.data, all_task_within_lobes); const within_filmLobeData = makeLobeData(task_film_data_lobes.within_corr.data, film_within_lobes); const within_restLobeData = makeLobeData(task_rest_data_lobes.within_corr.data, rest_within_lobes); display({within_allTaskLobeData_len: within_allTaskLobeData.length, within_filmLobeData_len: within_filmLobeData.length, within_restLobeData_len: within_restLobeData.length});*/49:T894, function getStats(values) { const sorted = values.slice().sort((a, b) => a - b); const n = sorted.length; const q1 = sorted[Math.floor(n * 0.25)]; const q3 = sorted[Math.floor(n * 0.75)]; const iqr = q3 - q1; const lowerFence = q1 - 1.5 * iqr; const upperFence = q3 + 1.5 * iqr; //Determine min and max values/upper and lower limit const withinFences = values.filter(v => v >= lowerFence && v <= upperFence); const min = withinFences.length > 0 ? Math.min(...withinFences) : q1; const max = withinFences.length > 0 ? Math.max(...withinFences) : q3; return { min: min, q1: q1, median: sorted[Math.floor(n * 0.5)], q3: q3, max: max, n: n, lowerFence: lowerFence, upperFence: upperFence }; } function make_Scatter(data, title, subtitle) { const typeOrder = ["within", "across"]; const statsData = typeOrder.flatMap(type => lobeOrder.map(region => { const values = data.filter(d => d.region === region && d.type === type).map(d => d.value); if (values.length === 0) return null; const stats = getStats(values); return { region: region, type: type, ...stats }; }).filter(d => d !== null) ); return Plot.plot({ width: 1500, height: 600, marginBottom: 60, x: { domain: lobeOrder, label: null, tickRotate: -30 }, y: { domain: [-1, 1], grid: true, label: "Correlation" }, title, subtitle, marks: [ Plot.ruleY([0], { stroke: "#999" }), Plot.dot(data, { x: "region", y: "value", fx: "type", fill: "region", r: 2, fillOpacity: 0.35, tip: true }), ] }); } const combinedAllTaskData = [ ...allTaskLobeData.map(d => ({ ...d, type: "across" })), ...within_allTaskLobeData.map(d => ({ ...d, type: "within" })) ]; const allTaskScatter = display(make_Scatter( combinedAllTaskData, "All Task — Within vs. Across-Model Reconstruction Correlation by Brain Region", `n = ${within_allTaskLobeData.length} (within) / ${allTaskLobeData.length} (across) locations` )); 4a:T981,function make_Scatter(data, title, subtitle) { const typeOrder = ["within", "across"]; const statsData = typeOrder.flatMap(type => lobeOrder.map(region => { const values = data.filter(d => d.region === region && d.type === type).map(d => d.value); if (values.length === 0) return null; const stats = getStats(values); return { region: region, type: type, ...stats }; }).filter(d => d !== null) ); return Plot.plot({ width: 1500, height: 600, marginBottom: 60, x: { domain: lobeOrder, label: null, tickRotate: -30 }, y: { domain: [-1, 1], grid: true, label: "Correlation" }, title, subtitle, marks: [ Plot.ruleY([0], { stroke: "#999" }), Plot.dot(data, { x: "region", y: "value", fx: "type", fill: "region", r: 2, fillOpacity: 0.35, tip: true }), Plot.boxY(data, { x: "region", y: "value", fx: "type", stroke: "region", strokeWidth: 1.5, fill: "region", stroke: "region", fillOpacity: 0.3 }), ] }); } // Code to Plot da plots const combinedAllTaskData = [ ...allTaskLobeData.map(d => ({ ...d, type: "across" })), ...within_allTaskLobeData.map(d => ({ ...d, type: "within" })) ]; const combinedRestData = [ ...restLobeData.map(d => ({ ...d, type: "across" })), ...within_restLobeData.map(d => ({ ...d, type: "within" })) ]; const combinedFilmData = [ ...filmLobeData.map(d => ({ ...d, type: "across" })), ...within_filmLobeData.map(d => ({ ...d, type: "within" })) ]; const allTaskScatter = display(make_Scatter( combinedAllTaskData, "All Task — Within vs. Across-Model Reconstruction Correlation by Brain Region", `n = ${within_allTaskLobeData.length} (within) / ${allTaskLobeData.length} (across) locations` )); const restScatter = display(make_Scatter( combinedRestData, "Rest Task — Within vs. Across-Model Reconstruction Correlation by Brain Region", `n = ${within_restLobeData.length} (within) / ${restLobeData.length} (across) locations` )); const filmScatter = display(make_Scatter( combinedFilmData, "Film Task — Within vs. Across-Model Reconstruction Correlation by Brain Region", `n = ${within_filmLobeData.length} (within) / ${filmLobeData.length} (across) locations` )); 4b:[{"id":"8ef3032677333b6813ac792b560c972a57fb404082fc19904ef2a967df88061ba06bb5766ca6bf3d9e9fd7fba0c7114a2b80cd5b277cb92f45444bce32d7a40b","href":"https://static.observableusercontent.com/files/8ef3032677333b6813ac792b560c972a57fb404082fc19904ef2a967df88061ba06bb5766ca6bf3d9e9fd7fba0c7114a2b80cd5b277cb92f45444bce32d7a40b","name":"dropped.png","type":"image/png","lastModified":1788643715019,"size":22720},{"id":"b4c90e6b684eb32ae9d929ae7fd22c34df6ae371e31c97ab5622e0b1944b228046fdabe971cae44f246993df09eb850771f6ceb113f219316221a1efd33f53f6","href":"https://static.observableusercontent.com/files/b4c90e6b684eb32ae9d929ae7fd22c34df6ae371e31c97ab5622e0b1944b228046fdabe971cae44f246993df09eb850771f6ceb113f219316221a1efd33f53f6","name":"all-task_data.npz","type":"application/octet-stream","lastModified":1788665989095,"size":399790},{"id":"d51d4813cab84f9feec812ad3dd88b2ce69b1e2343eb93dc8362683efcdce0a0dbe11b579459806830c129dcf944485cfc47fa9d9699555dc03bac49343747db","href":"https://static.observableusercontent.com/files/d51d4813cab84f9feec812ad3dd88b2ce69b1e2343eb93dc8362683efcdce0a0dbe11b579459806830c129dcf944485cfc47fa9d9699555dc03bac49343747db","name":"task-film_data.npz","type":"application/octet-stream","lastModified":1788665292037,"size":231982},{"id":"576a4293d4acc5873c7656f3f135ce929fdbf90ddb3cb9d449d6ab239c35209d135269736df8b117c77b70fdfeff6d9e2f46d81b5c7412b4a6611e776abd43b4","href":"https://static.observableusercontent.com/files/576a4293d4acc5873c7656f3f135ce929fdbf90ddb3cb9d449d6ab239c35209d135269736df8b117c77b70fdfeff6d9e2f46d81b5c7412b4a6611e776abd43b4","name":"task-rest_data.npz","type":"application/octet-stream","lastModified":1788665318140,"size":168910},{"id":"35e147f040bca73a82e0e7ce8afdbc93f3b639ae01f8db7651cd82b666a2386a41c6b2420097b941be48135be8fe573975501378bab57d69da0982b205bbd4b0","href":"https://static.observableusercontent.com/files/35e147f040bca73a82e0e7ce8afdbc93f3b639ae01f8db7651cd82b666a2386a41c6b2420097b941be48135be8fe573975501378bab57d69da0982b205bbd4b0","name":"all_across_lobe.json","type":"application/json","lastModified":1788838553368,"size":44429},{"id":"9b378db89d0b0ac6456c2c683196c9e27dfcc2cd33d382ffb33d9c921b8a3b41ba9079d75377e5c0e990c81403cadbcab6d0f89e7bbfc21eec557781129e05bb","href":"https://static.observableusercontent.com/files/9b378db89d0b0ac6456c2c683196c9e27dfcc2cd33d382ffb33d9c921b8a3b41ba9079d75377e5c0e990c81403cadbcab6d0f89e7bbfc21eec557781129e05bb","name":"all-task_data_lobes.npz","type":"application/octet-stream","lastModified":1788838546149,"size":55354},{"id":"f4c48a192591fcbfaea909b19cf968f8182e608d954729898a8e2095e092866b8663367240bd5bb8a9e38ff70441574f6979de9191bb1b1c683ebf62984093b1","href":"https://static.observableusercontent.com/files/f4c48a192591fcbfaea909b19cf968f8182e608d954729898a8e2095e092866b8663367240bd5bb8a9e38ff70441574f6979de9191bb1b1c683ebf62984093b1","name":"all_within_lobe.json","type":"application/json","lastModified":1788838563291,"size":44324},{"id":"b3425c58262e2e24e38ef3729c1a23dc60c26b15a0ef7d5c92dd808b69abd66ae1885565ec1001574fd9714d246d98ebc88e46bc59d3730cfb94a18363e02c8c","href":"https://static.observableusercontent.com/files/b3425c58262e2e24e38ef3729c1a23dc60c26b15a0ef7d5c92dd808b69abd66ae1885565ec1001574fd9714d246d98ebc88e46bc59d3730cfb94a18363e02c8c","name":"film_across_lobe.json","type":"application/json","lastModified":1788838478991,"size":25996},{"id":"2b6ce34f427fc7c8b8d5efa4d875736aa3faaa287471ff8a2033d8551886ba90ef898516de1e334ddf1f590c29ca2f6c4b2685e8fb0211d265a35cd88d42b352","href":"https://static.observableusercontent.com/files/2b6ce34f427fc7c8b8d5efa4d875736aa3faaa287471ff8a2033d8551886ba90ef898516de1e334ddf1f590c29ca2f6c4b2685e8fb0211d265a35cd88d42b352","name":"film_within_lobe.json","type":"application/json","lastModified":1788838460847,"size":25948},{"id":"29a236479058b88c5ecddb29c160443ca94f8400a8828872dc4537d92074d29a2d1dbdade9941ec5a5080965cef325be17eec40952c56bccff50942dc0904a3a","href":"https://static.observableusercontent.com/files/29a236479058b88c5ecddb29c160443ca94f8400a8828872dc4537d92074d29a2d1dbdade9941ec5a5080965cef325be17eec40952c56bccff50942dc0904a3a","name":"rest_across_lobe.json","type":"application/json","lastModified":1788838352447,"size":18433},{"id":"04a7e2cf99c76d1fa3e6c97763b9b6bbb01e7d8bd5ee4864a1fe537d861dc0bbe6a4fca96d64037425f1df08fa293b34b8ab5906f818d2038ccb6f880833e81b","href":"https://static.observableusercontent.com/files/04a7e2cf99c76d1fa3e6c97763b9b6bbb01e7d8bd5ee4864a1fe537d861dc0bbe6a4fca96d64037425f1df08fa293b34b8ab5906f818d2038ccb6f880833e81b","name":"rest_within_lobe.json","type":"application/json","lastModified":1788838391741,"size":18376},{"id":"dc9e89793195fa6fb4a254350bb8cea01f4aa3d2cf6511c7ce201aff912f92db801296d77916d79a0876e8b4f7eed2c02e515e72cd23a86df829d259bbc7aad4","href":"https://static.observableusercontent.com/files/dc9e89793195fa6fb4a254350bb8cea01f4aa3d2cf6511c7ce201aff912f92db801296d77916d79a0876e8b4f7eed2c02e515e72cd23a86df829d259bbc7aad4","name":"task-film_data_lobes.npz","type":"application/octet-stream","lastModified":1788838448303,"size":32786},{"id":"34211bad57a193422bc874bb3f41cfa4dbe134faa1b18ea3215546279f30e0648a1eb9c384981ff2e4fdefdb8f15df774d40afccdf174668f596e5433c4b57c2","href":"https://static.observableusercontent.com/files/34211bad57a193422bc874bb3f41cfa4dbe134faa1b18ea3215546279f30e0648a1eb9c384981ff2e4fdefdb8f15df774d40afccdf174668f596e5433c4b57c2","name":"task-rest_data_lobes.npz","type":"application/octet-stream","lastModified":1788838053145,"size":23098}] 39:{"id":"18ba0002ffb7a687","slug":"project-1-univariate-and-bivariate-data","title":"Project 1: Univariate and Bivariate Data","thumbnail":"888a3aa6882a50dff820a5984f32b74a70f88aee3564bce5f5e10ae982b15d01","access_level":"public","create_time":"$D2026-09-04T22:07:39.279Z","update_time":"$D2026-09-11T03:22:40.225Z","view_time":null,"first_public_time":"$D2026-09-09T06:34:41.910Z","has_unpublished_edits":null,"fork":false,"stars":0,"starred":false,"owner":{"type":"team","id":"6e1a0b6a59a98d7d","login":"jose-carmona-sanchez","name":"Jose.carmona-sanchez","avatar_url":"https://avatars-next.observableusercontent.com/7c7c926efd11da0bb4d75ebd1a0198795094c32af03dd95ecac6eebd7c4f7fb5","tier":"starter_2024","bio":"","following":false},"version":908,"publish_version":908,"role":null,"authored":false,"authors":[{"type":"individual","id":"ef578cb74e81811e","login":"jose-carmona-sanchez","name":"Jose.carmona-sanchez","avatar_url":"https://avatars-next.observableusercontent.com/7c7c926efd11da0bb4d75ebd1a0198795094c32af03dd95ecac6eebd7c4f7fb5","tier":null,"bio":"","following":false}],"members":[{"type":"team","id":"6e1a0b6a59a98d7d","login":"jose-carmona-sanchez","name":"Jose.carmona-sanchez","avatar_url":"https://avatars-next.observableusercontent.com/7c7c926efd11da0bb4d75ebd1a0198795094c32af03dd95ecac6eebd7c4f7fb5","tier":"starter_2024","role":"editor"}],"comments":[],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":0,"collection":null,"type":"notebook","body":{"titling":"auto","title":"Project 1: Univariate and Bivariate Data","cells":[{"id":97,"value":"# Project 1: Univariate and Bivariate Data","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":2,"value":"$3b","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":57,"value":"$3c","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":58,"value":"$3d","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":51,"value":"# A Gaussian Process Model of Human Electrophysiology Data with Varied Tasks\n### Jose Carmona","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":53,"value":"$3e","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":116,"value":"$3f","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":59,"value":"$40","pinned":false,"hidden":true,"mode":"js","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":96,"value":"### Results and Critiques: Distribution of Correlation Coefficients for Different Task Types\nAs seen above the hypothesis was wrong, more data will not always give a higher quantity of highly correlated data regardless of tasks. This entails that task type is important to consider if one wants to produce the best reconstructions with the GPR model. Some critiques that I do have about the graphs is that the data for across and within is positioned on top of each other on the histograms. I would much rather have the bars have some transparency so that you can see the true frequency. For astehtics I would also like each histogram to have its own y-axis so that there isn't as much open space above the rest and film data.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":120,"value":"$41","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":98,"value":"### Brain Region and GPR model Performance\nTo further investigate the influence that task data has on the GPR model I thought it would be interesting to see if certain brain regions would have better reconstruction/higher density of highly correlated coefficients.\n\nHere I test the hypothesis:\n\n**When comparing the GPR model reconstruction for the different task (rest vs film) the brain regions should differ in term of distribution of correlation coefficients.**\n","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":91,"value":"\n \n The specific brain region was extracted from the electrode coordinates. The correlation coefficient was then paired up with the respective electrode brain lobe and graphed as a scatter plot with a corresponding box plot overlapped. Each plot facets the within-subject model (left) against the across-subject model (right) for a given task dataset (All Task, Rest Task, Film Task).\n \n","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":99,"value":"$42","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":92,"value":"$43","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":100,"value":"### Sources\n- Original data used: https://openneuro.org/datasets/ds003688/versions/1.0.7\n- Jeremy Manning and Lucy Owen original paper on the GPR model: https://academic.oup.com/cercor/article/30/10/5333/5851264?__cf_chl_tk=Vs_JYH4EcOc2U5sum0KYfPVIE7jqZenRjb2JQDGtQiU-1765925269-1.0.1.1-o3YexL2QwzmC4KL2PrZtLyAzuh03DbacW.EXaAcY90E","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":102,"value":"## Revisions","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":111,"value":"With these specific results I already had in mind the plot that I wanted to do, it was a histogram that would be accompanied by a box plot where you can see all sub-datasets at once (without scrolling). So the first steps would be graphing a histogram and box plot which can be seen below. The toughest part was setting up my data in a way that was appropriate for the plots. I ended flattening my array and mapping it with what model type (within) and which dataset was being used (All task). This ended up makeing my life much easier in the final product when I wanted to combine model type and the different dataset.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":103,"value":"\nconst alltask_within_data = [\n ...Array.from(all_task_data.within_corr.data).map(v => ({ value: v, group: \"within\", dataset: \"All Task\" })),\n];\n\n\nconst corrDomain = [-1,1];\n\n\nconst color = {domain: [\"within\"],range: [\"#c24806\"]};\n\n\nconst histPlot = Plot.plot({\n width: 1000,\n height: 450,\n x: { domain: corrDomain, grid: true, label: \"Correlation\" },\n y: { grid: true, label: \"Frequency\", nice: true },\n color: { ...color, legend: true },\n marks: [\n Plot.rectY(\n alltask_within_data,\n Plot.binX({ y: \"count\" }, {\n x: \"value\",\n fill: \"group\",\n mixBlendMode: \"multiply\"\n })\n ),\n Plot.ruleY([0])\n ]\n});\n\n\ndisplay(html`\n

GPR Model Reconstruction Correlation Distribution for All Task (Within model)

\n ${histPlot}\n

GPR Model Reconstruction Correlation Distribution for All Task (Within model)

\n ${boxPlot}\n

\n`);","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":113,"value":"The next step was seeing how to combine the histogram and box plot. This was rather simple due to the html setup where you can just define the plots to put them together.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":101,"value":"$44","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":114,"value":"The next step was showcasing both model types within the same plots. I knew that all my data would have to be in a single array from online example so I applied the code I used above but with the across and within data for all task. It took some playing around to figure how to setup the settings so that labels and color would correctly be shown but I eventually got it.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":107,"value":"$45","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":115,"value":"The final step was to figure out how to showcase all the different datasets. Which ended being much more staright forward that I thought. The key to this part was finding out about the fx option within plot.plot which allows for subplots. I took the same philosophy with combining all the data together than it was just a matter of defining the different datasets/facets.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":117,"value":"$46","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":110,"value":"$47","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":122,"value":"For the scatter plot the first step was to pair up the correlation values and respective lobe region. For this data I imported two different types of array the correlation value and lobe region. Both arrays were had corresponding indexes so the correlation value for index 1 would correspond to the brain region in index 1. i ended up using the same .map feature to pair up the correlation value and brain region. The code below creates the arrays.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":121,"value":"$48","pinned":true,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":125,"value":"After I got the data, I wanted to test basic scatter plot capabilities and how to graph both within and across data. I used the same technique with the univariate graph where I pooled the within and acorss data together within one array where I flattened and mapped the the array. Using the same formatting as the univariate graph came up with the scatter plots below.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":119,"value":"$49","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":128,"value":"I thought that the scatter plots could be more informative so I went to add a box plot. Which was rather simple although the first implement had the box plot be filled grey but the fix was simple enough. The final step was to add sample count to each specific brain region and a upper and lower limit line to distinguish outliers which can be seen in the final graph.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"},{"id":126,"value":"$4a","pinned":false,"hidden":false,"mode":"ts","author":"$undefined","summary":"$undefined","tool":"$undefined","providerMetadata":"$undefined","attachments":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W4b","stdlib":"2"}} 3a:"$39:body"