31:T61c,task3_char1 = { //charts division basically for quantitative // init + query // Details of the experiment in the paper. describe methods such a way any grad student can replicate // software for running the benchmarking // Research contributions // 1. experiment results // 2. conclusions // 3. benchmarking framework const line = vl.markLine({ tooltip: true, point: true }).encode( vl .y() .fieldQ("filterTime") .aggregate("median") .scale({ type: "pow", exponent: "0.2" }) .title("Filter time(ms)"), vl.color().fieldN("library"), vl .x() .fieldO("experimentName") .title("record Count") .axis({ labelExpr: "datum.label + 'k'", labelAngle: "0" }) // vl // .row() // .fieldN("attrComb") // .title("Avg Filter Time") // .sort(["Categorical + Quantitative", "Quantitative", "Categorical"]), // vl // .column() // .fieldN("valComb") // .title("values selected") // .sort([ // "No", // "Random", // "All", // "Random-No", // "All-No", // "All-Random", // "All-No-Random" // ]) ); const band = line.markErrorband({ extent: "iqr" }); return vl .layer(band, line) .height(100) .width(300) .facet({ column: vl.fieldN("Selected Values"), row: vl.fieldN("Attribute Type") }) .data( transformData({ data: task2 }).filter((d) => ["All", "Random", "No"].includes(d["Selected Values"]) ) ) .render(); }32:T620,task3_chart2 = { //charts division basically for quantitative // init + query // Details of the experiment in the paper. describe methods such a way any grad student can replicate // software for running the benchmarking // Research contributions // 1. experiment results // 2. conclusions // 3. benchmarking framework const line = vl.markLine({ tooltip: true, point: true }).encode( vl .y() .fieldQ("filterTime") .aggregate("median") .scale({ type: "pow", exponent: "0.2" }) .title("Filter time(ms)"), vl.color().fieldN("library"), vl .x() .fieldO("experimentName") .title("record Count") .axis({ labelExpr: "datum.label + 'k'", labelAngle: "0" }) // vl // .row() // .fieldN("attrComb") // .title("Avg Filter Time") // .sort(["Categorical + Quantitative", "Quantitative", "Categorical"]), // vl // .column() // .fieldN("valComb") // .title("values selected") // .sort([ // "No", // "Random", // "All", // "Random-No", // "All-No", // "All-Random", // "All-No-Random" // ]) ); const band = line.markErrorband({ extent: "iqr" }); return vl .layer(band, line) .height(100) .width(200) .facet({ column: vl.fieldN("Selected Values"), row: vl.fieldN("Attribute Type") }) .data( transformData({ data: task2 }).filter( (d) => !["All", "Random", "No"].includes(d["Selected Values"]) ) ) .render(); }33:T485,barChart = { let target = []; for (let data of task2) { let res = {}; res["comb"] = data.comb; res["duckdb"] = data["duckDB-filter_time"]; res["arquero"] = data["arquero-filter_time"]; res["nativeJS"] = data["vanillaJS-filter_time"]; res["dataset"] = data.dataset; target.push(res); } let data = []; if (target.length) { data = aq .from(target) .fold(["duckdb", "arquero", "nativeJS"], { as: ["backend", "time"] }) .objects(); } const label = vl .markText({ dx: 5, dy: 5, align: "start", fontSize: "8" }) .encode( vl.y().fieldN("backend").title("Filtering Operation"), vl.x().median("time").title("Median Time with IQR"), vl.text().median("time").format(".4f") ) .height(100) .width(100); const bar = label .markBar({ tooltip: true, opacity: 0.6 }) .encode(vl.color().fieldN("backend").title("Filtering Operation")) .height(100) .width(100); const error = bar.markErrorbar({ extent: "iqr" }); return vl .layer(label, bar, error) .facet(vl.fieldN("comb")) .spacing(15) .columns(6) .data(data) .render(); }34:T40a,async function displayPlot(data) { //statistacially difference // order by blank charts. // categorical // Why do have u //Report -> worker -> // landing notebook -> results and then details of the experiment const line = vl .markLine({ tooltip: true, point: true }) .encode( vl .y() .fieldQ("filterTime") .aggregate("median") .scale({ type: "pow", exponent: "0.2" }) .title("Filter Time (ms)"), vl.color().fieldN("library"), vl .x() .fieldO("experimentName") .title("record Count") .axis({ labelExpr: "datum.label + 'k'", labelAngle: "0" }) ); const band = line.markErrorband({ extent: "iqr" }); let chart = await vl .layer(band, line) .height(170) .width(700) .facet({ column: vl.fieldN("dataset") // row: vl.fieldN("Attribute Type") }) .data( transformData({ data: data }).filter((d) => d["experimentName"] < 600) ) .render(); return htl.html`${chart}`; }35:T440,function transformData({ data }) { if (!data || data.length === 0) { return []; } let removeColumns = Object.keys(data[0]).reduce((prev, d) => { d.includes("-") && prev.push(d); return prev; }, []); let tempData = aq .from(data) .fold( ["vanillaJS-filter_time", "arquero-filter_time", "duckDB-filter_time"], { as: ["library", "filterTime"] } ) .derive({ filterTime: (d) => +d.filterTime, library: (d) => op.split(d["library"], "-")[0], experimentName: (d) => +op.split(d["experimentName"], "_")[2] / 1000, initTime: aq.escape((d) => d[`${d["library"]}-init_time`]), "Attribute Type": aq.escape((d) => checkAttributes(d["comb"])), "Selected Values": aq.escape((d) => checkValues(d["comb"])) }) .select(aq.not(removeColumns)); return tempData.objects().map((d) => { d["library"] === "duckDB" ? (d["library"] = "DuckDB-WASM") : ""; d["library"] === "arquero" ? (d["library"] = "Arquero") : "", d["library"] === "vanillaJS" ? (d["library"] = "VanillaJS") : ""; return d; }); }36:[{"id":"239044d710076820748b4f175c8285fadedc5367de5de14264e39bf4faa7510a7d24e0b0e32b663433318216458b5b4c2a7d690349ee6d67d20619b1e6bd5ab0","href":"https://static.observableusercontent.com/files/239044d710076820748b4f175c8285fadedc5367de5de14264e39bf4faa7510a7d24e0b0e32b663433318216458b5b4c2a7d690349ee6d67d20619b1e6bd5ab0","name":"vivek_system.json","type":"application/json","lastModified":1702572953251.29,"size":72309},{"id":"bfc379aec6fecd5b0d86180f94619cb975e3f8195d5142a0980cd2df5d479f792fe2ac17c1e270f4d042cb50969ad9098a529b3033ca45ffbd9810f3c053ae21","href":"https://static.observableusercontent.com/files/bfc379aec6fecd5b0d86180f94619cb975e3f8195d5142a0980cd2df5d479f792fe2ac17c1e270f4d042cb50969ad9098a529b3033ca45ffbd9810f3c053ae21","name":"vivek_system_specs.png","type":"image/png","lastModified":1702573745499.134,"size":30474},{"id":"811ddae4b64bc58709f188bad8b4ac8ecfefe2102a67066bb6b0dfae99516627679808cd9d57682ad4a33f957cd12c3f8a433842b9a2704880d0f37928151489","href":"https://static.observableusercontent.com/files/811ddae4b64bc58709f188bad8b4ac8ecfefe2102a67066bb6b0dfae99516627679808cd9d57682ad4a33f957cd12c3f8a433842b9a2704880d0f37928151489","name":"akhila_mac_specs.png","type":"image/png","lastModified":1702573716125.259,"size":28315},{"id":"040ee369b44802fe3bc759a0a061d2138efefe28441df90315f94cd678eaf0e740b660ca396e544cc2c83186cb8e2203952e8d822c94306ae2b492f75eb5954b","href":"https://static.observableusercontent.com/files/040ee369b44802fe3bc759a0a061d2138efefe28441df90315f94cd678eaf0e740b660ca396e544cc2c83186cb8e2203952e8d822c94306ae2b492f75eb5954b","name":"harshit_mac_spec.png","type":"image/png","lastModified":1702573920804.654,"size":28118},{"id":"3e8ccbed37eeae7cb8db1ad264fd64f4c092e58a54831026136ac8edc3fd8b1652fdeca63a15421ed275b480dcccda0e01f7d66b999df7e06bf2c9d201eee6d8","href":"https://static.observableusercontent.com/files/3e8ccbed37eeae7cb8db1ad264fd64f4c092e58a54831026136ac8edc3fd8b1652fdeca63a15421ed275b480dcccda0e01f7d66b999df7e06bf2c9d201eee6d8","name":"task1.json","type":"application/json","lastModified":1702658530959.177,"size":220762},{"id":"88cc394af0179b58e5ed929ab6c2d46cbbdfc3a85ed4a1ae19c937e7b4692f5d94a7a967ff89b06a813535309613a6e8570a6afb204cec1eaf71b9df150b3437","href":"https://static.observableusercontent.com/files/88cc394af0179b58e5ed929ab6c2d46cbbdfc3a85ed4a1ae19c937e7b4692f5d94a7a967ff89b06a813535309613a6e8570a6afb204cec1eaf71b9df150b3437","name":"task2.json","type":"application/json","lastModified":1702659182646.526,"size":1773109},{"id":"a751a2cb40331555d35a4d6c20c43d61fe29a7b0c4986636a98f4a4b933e83727d22dd2ca308add47a8b8f019db51ad048d466c84f8d843e0838d81e51a63d59","href":"https://static.observableusercontent.com/files/a751a2cb40331555d35a4d6c20c43d61fe29a7b0c4986636a98f4a4b933e83727d22dd2ca308add47a8b8f019db51ad048d466c84f8d843e0838d81e51a63d59","name":"harshit_system.json","type":"application/json","lastModified":1702540883617.735,"size":73749},{"id":"f1309f19b06b7a69936ffab1fcb7799a11a3bf1abfa85750baa93f474ce98cb4e40d3940bec7a34c3c9758039441a9526275fb03d3d32678b051dc5a681f9210","href":"https://static.observableusercontent.com/files/f1309f19b06b7a69936ffab1fcb7799a11a3bf1abfa85750baa93f474ce98cb4e40d3940bec7a34c3c9758039441a9526275fb03d3d32678b051dc5a681f9210","name":"akhila_system.json","type":"application/json","lastModified":1702540925576.41,"size":41948}] 2f:{"title":"Benchmarking Results","cells":[{"id":58,"value":"# Benchmarking Results ","pinned":false,"hidden":false,"mode":"md"},{"id":193,"value":"## Task1\nIdentify the most efficient in-browser data processing library for data filtering operations with a record count of 100k, we have created an experiment that generated 560 queries on 4 datasets Moma, Flights, Weather, and IMDB.","pinned":false,"hidden":false,"mode":"md"},{"id":197,"value":"vl\n .markBoxplot({ tooltip: true })\n .encode(\n vl\n .x()\n .fieldQ(\"filterTime\")\n .scale({ type: \"pow\", exponent: 0.2 })\n .title(\"Filter Time(in ms)\"),\n vl.y().fieldN(\"library\"),\n vl.color().fieldN(\"library\")\n )\n .data(transformData({ data: task1 }))\n .width(800)\n .height(100)\n .render()","pinned":false,"hidden":false,"mode":"ojs"},{"id":205,"value":"VanillaJS exhibits a lower median filter time compared to Arquero and DuckDB-WASM for a record count of 100,000 across various datasets. Additionally, the graph indicates that for certain queries, both Arquero and VanillaJS take over 6 seconds to execute, whereas DuckDB-WASM demonstrates a consistent execution time of approximately 1 second","pinned":false,"hidden":false,"mode":"md"},{"id":200,"value":"task1 = FileAttachment(\"task1.json\").json()","pinned":false,"hidden":false,"mode":"ojs"},{"id":214,"value":"## Task 2\nCompare the execution time performance for data filtering tasks using Arquero,\nDuckDB-WASM, and VanillaJS for Categorical and Quantitative attribute types, and observe\ntheir trends across a range of record sizes: 0.5k, 5k, 10k, 20k, 30k, 40k, and 50k.","pinned":false,"hidden":false,"mode":"md"},{"id":224,"value":"{\n //statistacially difference\n // order by blank charts.\n // categorical\n // Why do have u\n //Report -> worker ->\n // landing notebook -> results and then details of the experiment\n const line = vl\n .markLine({ tooltip: true, point: true })\n .encode(\n vl\n .y()\n .fieldQ(\"filterTime\")\n .aggregate(\"median\")\n .scale({ type: \"pow\", exponent: \"0.2\" })\n .title(\"Filter Time (ms)\"),\n vl.color().fieldN(\"library\"),\n vl\n .x()\n .fieldO(\"experimentName\")\n .title(\"record Count\")\n .axis({ labelExpr: \"datum.label + 'k'\", labelAngle: \"0\" })\n );\n\n const band = line.markErrorband({ extent: \"iqr\" });\n\n return vl\n .layer(band, line)\n .height(100)\n .width(200)\n .facet({\n column: vl.fieldN(\"dataset\"),\n row: vl.fieldN(\"Attribute Type\")\n })\n .data(transformData({ data: task2 }))\n\n .render();\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":237,"value":"It is evident that for record counts under 50,000, VanillaJS consistently maintains a less median filter time than Arquero and DuckDB-WASM, irrespective of the type of attribute or the dataset used.","pinned":false,"hidden":false,"mode":"md"},{"id":227,"value":"task2 = await FileAttachment(\"task2.json\").json()","pinned":true,"hidden":false,"mode":"ojs"},{"id":242,"value":"## Task 3\nCompare the execution time performance for data filtering tasks using Arquero,\nDuckDB-WASM, and VanillaJS for Categorical and Quantitative attribute types with differ-\nent values selected, and observe their trends across a range of record sizes: 0.5k, 5k, 10k, 20k,\n30k, 40k, and 50k","pinned":false,"hidden":false,"mode":"md"},{"id":249,"value":"$31","pinned":false,"hidden":false,"mode":"ojs"},{"id":244,"value":"$32","pinned":false,"hidden":false,"mode":"ojs"},{"id":256,"value":" VanillaJS generally outperforms Arquero and DuckDB-WASM in terms of speed in most scenarios, even with varying\nselected values. The exception is observed when dealing with both categorical and quantita-\ntive data, where DuckDB-WASM exhibits faster perform","pinned":false,"hidden":false,"mode":"md"},{"id":220,"value":"## Task 4\ncompare the execution efficiency in terms of data filtering tasks among Arquero,\nDuckDB-WASM, and VanillaJS across different systems, and observe their trends for pro-\ncessing record counts up to 500,000.","pinned":false,"hidden":false,"mode":"md"},{"id":49,"value":"htl.html`

System 1 Results ${exp1_chart} System 2 Results ${exp2_chart} System 3 Results ${exp3_chart}

`","pinned":false,"hidden":false,"mode":"ojs"},{"id":273,"value":"t2 = task2.map((d) => {\n d[\"query combination\"] = d[\"comb\"];\n return d;\n})","pinned":true,"hidden":false,"mode":"ojs"},{"id":337,"value":"

Benchmarking Data Filtering operation

","pinned":false,"hidden":false,"mode":"md"},{"id":271,"value":"$33","pinned":true,"hidden":false,"mode":"ojs"},{"id":163,"value":"## Experiment Details","pinned":false,"hidden":false,"mode":"md"},{"id":126,"value":"Inputs.table(combined_experiments, {\n header: [\n \"Record count\",\n \"Akhila System (success / total)\",\n \"Harshit System (success / total)\",\n \"Vivek System (success / total)\"\n ]\n})","pinned":false,"hidden":false,"mode":"ojs"},{"id":8,"value":"akhila_system = FileAttachment(\"akhila_system.json\").json()","pinned":false,"hidden":false,"mode":"ojs"},{"id":39,"value":"exp1_chart = htl.html`${await displayPlot(akhila_system)}`","pinned":false,"hidden":false,"mode":"ojs"},{"id":41,"value":"exp2_chart = displayPlot(harshit_system)","pinned":false,"hidden":false,"mode":"ojs"},{"id":7,"value":"harshit_system = FileAttachment(\"harshit_system.json\").json()","pinned":false,"hidden":false,"mode":"ojs"},{"id":66,"value":"viviek_system = FileAttachment(\"vivek_system.json\").json()","pinned":false,"hidden":false,"mode":"ojs"},{"id":68,"value":"exp3_chart = displayPlot(viviek_system)","pinned":false,"hidden":false,"mode":"ojs"},{"id":21,"value":"$34","pinned":false,"hidden":false,"mode":"ojs"},{"id":88,"value":"akhila_mac_specs = FileAttachment(\"akhila_mac_specs.png\").image()","pinned":false,"hidden":false,"mode":"ojs"},{"id":183,"value":"## System Specs","pinned":false,"hidden":false,"mode":"md"},{"id":89,"value":"harshit_mac_spec = FileAttachment(\"harshit_mac_spec.png\").image()","pinned":false,"hidden":false,"mode":"ojs"},{"id":90,"value":"vivek_system_specs = FileAttachment(\"vivek_system_specs.png\").image()","pinned":false,"hidden":false,"mode":"ojs"},{"id":117,"value":"combined_experiments = {\n let res = [];\n let keys = Array.from(akhila_experiments_details.keys()).sort(\n (a, b) => a - b\n );\n for (let exp of keys) {\n let ak = akhila_experiments_details.get(exp);\n let h = harshit_experiments_details.get(exp);\n let v = vivek_experiments_details.get(exp);\n res.push([\n \"\" + exp + \"K\",\n +ak > 10 ? ak + \" / 21\" : \"0\" + ak + \" / 21\",\n +h > 10 ? h + \" / 21\" : \"0\" + h + \" / 21\",\n +v > 10 ? v + \" / 21\" : \"0\" + v + \" / 21\"\n ]);\n }\n\n return res;\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":96,"value":"akhila_experiments_details = d3.rollup(\n transformData({ data: akhila_system }).filter(\n (d) => d[\"experimentName\"] < 600\n ),\n (v) => v.length / 3,\n (d) => d.experimentName\n)","pinned":false,"hidden":false,"mode":"ojs"},{"id":107,"value":"harshit_experiments_details = d3.rollup(\n transformData({ data: harshit_system }).filter(\n (d) => d[\"experimentName\"] < 600\n ),\n (v) => v.length / 3,\n (d) => d.experimentName\n)","pinned":false,"hidden":false,"mode":"ojs"},{"id":110,"value":"vivek_experiments_details = d3.rollup(\n transformData({ data: viviek_system }).filter(\n (d) => d[\"experimentName\"] < 600\n ),\n (v) => v.length / 3,\n (d) => d.experimentName\n)","pinned":false,"hidden":false,"mode":"ojs"},{"id":101,"value":"transformData({ data: akhila_system }).filter((d) => d[\"experimentName\"] < 600)","pinned":true,"hidden":false,"mode":"ojs"},{"id":23,"value":"$35","pinned":false,"hidden":false,"mode":"ojs"},{"id":26,"value":"function checkAttributes(comb) {\n const categoricalRegex = /C+/gi,\n quantitativeRegex = /Q+/gi;\n if (comb.match(categoricalRegex) && comb.match(quantitativeRegex)) {\n return \"Categorical + Quantitative\";\n }\n if (comb.match(categoricalRegex)) {\n return \"Categorical\";\n }\n if (comb.match(quantitativeRegex)) {\n return \"Quantitative\";\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":28,"value":"function checkValues(comb) {\n const allValuesRegex = /@+/g,\n noValuesRegex = /N+/gi,\n randomValuesRegex = /R+/gi;\n\n if (\n comb.match(allValuesRegex) &&\n comb.match(noValuesRegex) &&\n comb.match(randomValuesRegex)\n ) {\n return \"All-No-Random\";\n }\n if (comb.match(allValuesRegex) && comb.match(randomValuesRegex)) {\n return \"All-Random\";\n }\n if (comb.match(allValuesRegex) && comb.match(noValuesRegex)) {\n return \"All-No\";\n }\n if (comb.match(noValuesRegex) && comb.match(randomValuesRegex)) {\n return \"Random-No\";\n }\n\n if (comb.match(allValuesRegex)) {\n return \"All\";\n }\n\n if (comb.match(randomValuesRegex)) {\n return \"Random\";\n }\n\n if (comb.match(noValuesRegex)) {\n return \"No\";\n }\n}","pinned":false,"hidden":false,"mode":"ojs"},{"id":31,"value":"import { aq, op } from \"@uwdata/arquero\"","pinned":false,"hidden":false,"mode":"ojs"}],"files":"$W36","stdlib":"1","titling":"auto"}