2d:null 2e:["$","title",null,{"children":"Are LLM's capable of aiding data transformation? | Kasi Viswanath | Observable"}] 33:Tbbe,function formatTableRow( data, { formatColumns = {}, colorScheme = d3.schemeRdYlGn[3], columns = data.length != 0 ? Object.keys(data[0]) : [], //Table CSS Properties tableCSS = { "overflow-x": "auto", display: "block", "min-width": "100%", height: "100%", "max-height": "240px" }, // Function to compute the comparison value cmpValueFn = (d) => d } = {} ) { if (data.length === 0) { return; } let formatCSS = (d) => d && JSON.stringify(d).slice(1, -1).replaceAll('"', "").replaceAll(",", ";"); let formatColsKeys = Object.keys(formatColumns); data = data.map((d) => columns.reduce((prev, col) => { prev[col] = d[col]; return prev; }, {}) ); const table = d3 .select("body") .append("table") .property("style", formatCSS(tableCSS)); table .append("thead") .append("tr") .attr("class", "header") .selectAll("th") .data(columns) .join((enter) => enter.append("th").text((d) => d)); table .append("tbody") .selectAll("tr") .data(data) .join((enter) => enter.append("tr")) .attr("class", "row") .selectAll("td") .data((row) => { let cmpValue_ = cmpValueFn(row); return columns.map((col) => { let res = {}; res["value"] = row[col]; res["cmpValue_"] = cmpValue_; if (formatColsKeys.includes(col)) { let { condition, trueFormatting, falseFormatting } = formatColumns[col]; if (condition(res["cmpValue_"], row)) { res["style"] = trueFormatting; } else { res["style"] = falseFormatting; } } return res; }); }) .join((enter) => enter .append("td") .text((d) => d.value) .property("style", (d) => formatCSS(d.style)) ); function addColumn(col, data) { table .select(".header") .data([col]) .append("th") .text((d) => d); table .selectAll(".row") .data(data) .append("td") .transition() .delay((_, i) => i * 1000) .text((d) => d); } let btn = Inputs.button("Create column"); let newCol = Inputs.text({ label: "Column Name", placeholder: "country" }); let downloadFile = downloadbuttonCSV(data); let text = Inputs.textarea({ placeholder: "find capital city for ${row.country}" }); btn.addEventListener("click", async () => { let result = []; for (let row of data) { let res = await getLLMResponse(eval("`" + text.value + "`")); row[newCol.value] = res; result.push(res); console.log(res); } // data = result; addColumn( newCol.value, // result.map((d) => d.target) result ); text.value = ""; newCol.value = ""; }); let target = html`${table.node()}

${newCol} ${btn}

${text} ${downloadFile}`; target.data = data; return target; }34:[{"id":"4af1f2a8420ab01de9f5f51354c4d637756c36fc0a05cff35e94de457b3647ffa33e1212e165b28753cc7fd84cf4a48c87d8a7505944982346338e53cf1200dd","href":"https://static.observableusercontent.com/files/4af1f2a8420ab01de9f5f51354c4d637756c36fc0a05cff35e94de457b3647ffa33e1212e165b28753cc7fd84cf4a48c87d8a7505944982346338e53cf1200dd","name":"country-by-capital-city.json","type":"application/json","lastModified":1699825753218.119,"size":17937},{"id":"b44e97b013e138390fdac7ee23fa365520e82402ea658b80dadf24e443d6c6389639e6ed8e47a438c1627a199ac68b207fff6990751398aaad24c9fbb459dfb4","href":"https://static.observableusercontent.com/files/b44e97b013e138390fdac7ee23fa365520e82402ea658b80dadf24e443d6c6389639e6ed8e47a438c1627a199ac68b207fff6990751398aaad24c9fbb459dfb4","name":"benchmark_results.csv","type":"text/csv","lastModified":1698683557458.382,"size":690310},{"id":"9aadb14f2350a648b3747e5e3433421cd66dfc0decb84f7facdc0691176cd997313d8502964a4890ffdb5efa3b7fd72bec78034acc632b435b62fd59513c59f2","href":"https://static.observableusercontent.com/files/9aadb14f2350a648b3747e5e3433421cd66dfc0decb84f7facdc0691176cd997313d8502964a4890ffdb5efa3b7fd72bec78034acc632b435b62fd59513c59f2","name":"benchmark_results_10_replica_till_100k.csv","type":"text/csv","lastModified":1698684040710.252,"size":810}] 31:{"id":"3287edd5f283b698","slug":"transform-data-with-llm","title":"Are LLM's capable of aiding data transformation?","thumbnail":"87f5b13c0ce9e555abe23ebbdd36ebe5ae74c4b31487d2c51eb8b30b97979455","access_level":"public","create_time":"$D2023-10-24T03:09:25.735Z","update_time":"$D2023-12-24T23:49:37.684Z","view_time":null,"first_public_time":"$D2023-10-24T03:09:25.735Z","has_unpublished_edits":null,"fork":false,"stars":2,"starred":false,"owner":{"type":"team","id":"b760ea566a896d7d","login":"kasivisu4","name":"Kasi Viswanath","avatar_url":"https://avatars-next.observableusercontent.com/48e36723913e7e5b18b906e1d7cde12fad4146715e78bbf74d3db7969a248b56","tier":"starter_2024"},"version":864,"publish_version":null,"role":null,"authored":false,"authors":[{"type":"individual","id":"fcd94e1c1ca12ae9","login":"kasivisu4","name":"Kasi Viswanath","avatar_url":"https://avatars-next.observableusercontent.com/48e36723913e7e5b18b906e1d7cde12fad4146715e78bbf74d3db7969a248b56","tier":null}],"members":[{"type":"individual","id":"fcd94e1c1ca12ae9","login":"kasivisu4","name":"Kasi Viswanath","avatar_url":"https://avatars-next.observableusercontent.com/48e36723913e7e5b18b906e1d7cde12fad4146715e78bbf74d3db7969a248b56","tier":null,"role":"editor"},{"type":"individual","id":"be0076fec73aa550","login":"john-guerra","name":"John Alexis Guerra Gómez","avatar_url":"https://avatars-next.observableusercontent.com/21638cee44bb0ea70798da725b57111a5d3e2acebcd14524565992f7d1594aed","tier":null,"role":"viewer"}],"comments":[],"custom_thumbnail":false,"trashed":false,"lapsed":null,"subscribed":null,"forks":1,"collection":null,"type":"notebook","body":{"title":"Are LLM's capable of aiding data transformation?","cells":[{"id":0,"value":"# Are LLM's capable of aiding data transformation?\n\n```js\nimport {formatTableRow} from \"@kasivisu4/transform-data-with-llm\"\n```\n\nEither in Data Analysis or Data Processing, transforming data will help from converting raw data to required format.\n\nThe benefit of utilizing a language model for data transformation is that it's not just restricted to computational tasks; it also applies the model's extensive knowledge to derive values.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":812,"value":"We mainly utilize Google Bard's Large Language Models (since they are free of charge) for checking.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":502,"value":"viewof BARD_API_KEY = Inputs.text({\n label: \"BARD API KEY\",\n value: localStorage.getItem(\"BARD_API_KEY\"),\n placeholder: \"Enter the BARD API KEY\"\n})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":658,"value":"Before going through following examples get Google Bard API steps ","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":693,"value":"### Note\n- As Transformation happens in every row, it is essential to specify the columns you intend to utilize in the embedded expression. To reference the data in a row, simply use the term 'row'.\n - For instance, in example 1, if you need to access the 'country' column, you would write: \\${row.country}.\n- The API is limited to bard limit. For more information check here\n\n","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":386,"value":"#### Example 1 - Find Capital for the country","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":613,"value":"{\n let target = formatTableRow(data, {});\n target.querySelector(\"input\").value = \"capital\";\n target\n .querySelector(\"input\")\n .dispatchEvent(new Event(\"input\", { bubbles: true }));\n\n target.querySelector(\"textarea\").value =\n \"find capital city for ${row.country}\";\n target\n .querySelector(\"textarea\")\n .dispatchEvent(new Event(\"input\", { bubbles: true }));\n\n return target;\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":687,"value":"#### Example 2 - Find address in safe neighborhood or not?","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":484,"value":"{\n let target = formatTableRow(apts.slice(0, 10), {\n columns: [\"price\", \"bhk\", \"address\"]\n });\n target.querySelector(\"input\").value = \"safe_neighborhood\";\n target\n .querySelector(\"input\")\n .dispatchEvent(new Event(\"input\", { bubbles: true }));\n\n target.querySelector(\"textarea\").value =\n \"Is this address in ${row.address} safe or not?\";\n target\n .querySelector(\"textarea\")\n .dispatchEvent(new Event(\"input\", { bubbles: true }));\n\n return target;\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":716,"value":"Why not Chatgpt?\n\nIn my view, while ChatGPT often provides more precise responses, particularly with historical data, the Bard API holds the benefit of being free and offers smooth integration with Google services, including the ability to obtain location details, which is a feature less comprehensively available in ChatGPT.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":723,"value":"#### Example 3 - Find distance from my Northeastern university, San Francsico (600 California Street) to all apartments","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":731,"value":"ex3 = {\n let target = formatTableRow(apts.slice(0, 10), {\n columns: [\"price\", \"bhk\", \"address\"]\n });\n target.querySelector(\"input\").value = \"Distance\";\n target\n .querySelector(\"input\")\n .dispatchEvent(new Event(\"input\", { bubbles: true }));\n\n target.querySelector(\"textarea\").value =\n \"Find distance in miles from 600 California St, san franisco to ${row.address}?\";\n target\n .querySelector(\"textarea\")\n .dispatchEvent(new Event(\"input\", { bubbles: true }));\n\n return target;\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":777,"value":"### Try with your own data (CSV Supported)\nPlease note that if you have a dataset of 1000 records, each new column will trigger 1000 calls to the BARD API to generate the corresponding values.","pinned":false,"hidden":false,"mode":"md","author":"$undefined","summary":"$undefined","output":"","database":"$undefined"},{"id":774,"value":"viewof customData = dataInput()","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":785,"value":"formatTableRow(customData, {})","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":7,"value":"$33","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":861,"value":"new Set(penguins.map((d) => d.island))","pinned":true,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":465,"value":"countryByCapitalCity = FileAttachment(\"country-by-capital-city.json\").json()","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":285,"value":"benchmark_results = FileAttachment(\n \"benchmark_results_10_replica_till_100k.csv\"\n).csv({ typed: true })","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":519,"value":"data = [{ country: \"India\" }, { country: \"USA\" }]","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":589,"value":"import { data as apts } from \"@kasivisu4/craigslist-apt-data\"","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":480,"value":"async function getLLMResponse(question, API_KEY = BARD_API_KEY) {\n let myHeaders = new Headers();\n myHeaders.append(\"Content-Type\", \"application/json\");\n\n let formatQuestion = `${question} `;\n console.log({ formatQuestion });\n let raw = JSON.stringify({\n prompt: {\n text: formatQuestion.replace('\"', \"\")\n },\n temperature: 1\n });\n\n let requestOptions = {\n method: \"POST\",\n headers: myHeaders,\n body: raw,\n redirect: \"follow\"\n };\n\n let req = await fetch(\n `https://generativelanguage.googleapis.com/v1beta3/models/text-bison-001:generateText?key=${API_KEY}`,\n requestOptions\n );\n\n let res = await req.text();\n\n let target = \"\";\n\n try {\n target = JSON.parse(res)?.candidates[0]?.output;\n console.log(JSON.parse(res)?.candidates[0]);\n } catch (err) {\n console.error(err);\n } finally {\n return target;\n }\n}","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":755,"value":"import { downloadbuttonCSV } from \"@pavlos/download-buttons\"","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"},{"id":771,"value":"import { dataInput } from \"@john-guerra/data-input\"","pinned":false,"hidden":false,"mode":"ojs","author":"$undefined","summary":"$undefined","output":"$undefined","database":"$undefined"}],"files":"$W34","stdlib":"1","titling":"auto"}} 32:"$31:body"