Sarveni/Top-VS-Code-Extensions-Data-Analysis

1import fs from 'fs';2import Papa from 'papaparse';3 4const csvData = fs.readFileSync('./data.csv', 'utf8');5 6Papa.parse(csvData, {7    header: true,8    dynamicTyping: true,9    skipEmptyLines: true,10    complete: (results) => {11        const data = results.data;12        13        let validExtensions = [];14        let totalDownloads = 0;15        16        for (const row of data) {17            const dl = row.downloads_count;18            if (typeof dl !== 'number' || isNaN(dl) || dl < 0) {19                continue;20            }21            validExtensions.push(dl);22            totalDownloads += dl;23        }24        25        validExtensions.sort((a, b) => b - a);26        const totalExts = validExtensions.length;27 28        function getShare(fraction) {29            const count = Math.max(1, Math.floor(totalExts * fraction));30            let sum = 0;31            for(let i = 0; i < count; i++) sum += validExtensions[i];32            return { count, share: (sum / totalDownloads * 100).toFixed(2) };33        }34 35        function getCountForThreshold(threshold, extArray, totalDl) {36            let running = 0;37            for(let i = 0; i < extArray.length; i++) {38                running += extArray[i];39                if (running >= totalDl * threshold) return i + 1;40            }41            return extArray.length;42        }43 44        const val1_1 = getShare(0.01);45        const val1_5 = getShare(0.05);46        const val1_10 = getShare(0.10);47 48        const val2_50 = getCountForThreshold(0.50, validExtensions, totalDownloads);49        const val3_80 = getCountForThreshold(0.80, validExtensions, totalDownloads);50 51        const stab_05 = getShare(0.005);52        const stab_1 = getShare(0.01);53        const stab_2 = getShare(0.02);54        const stab_5 = getShare(0.05);55 56        // Outlier Test57        const noOutlierExts = validExtensions.slice(1);58        const noOutlierTotalDl = totalDownloads - validExtensions[0];59        const noOutlier_50 = getCountForThreshold(0.50, noOutlierExts, noOutlierTotalDl);60        const noOutlier_80 = getCountForThreshold(0.80, noOutlierExts, noOutlierTotalDl);61        62        const noOutlierTop1Count = Math.max(1, Math.floor(noOutlierExts.length * 0.01));63        let noOutlierTop1Sum = 0;64        for(let i=0; i < noOutlierTop1Count; i++) noOutlierTop1Sum += noOutlierExts[i];65        const noOutlierTop1Share = (noOutlierTop1Sum / noOutlierTotalDl * 100).toFixed(2);66 67        // Option A group calculation68        const top1Count = val1_1.count;69        const top5Count = val1_5.count;70        const top20Count = Math.max(1, Math.floor(totalExts * 0.20));71 72        let top1Sum = 0, next4Sum = 0, next15Sum = 0, bottom80Sum = 0;73        for (let i = 0; i < totalExts; i++) {74            const dl = validExtensions[i];75            if (i < top1Count) top1Sum += dl;76            else if (i < top5Count) next4Sum += dl;77            else if (i < top20Count) next15Sum += dl;78            else bottom80Sum += dl;79        }80 81        console.log(JSON.stringify({82            totalExts,83            totalDownloads,84            validation1: { top1: val1_1, top5: val1_5, top10: val1_10 },85            validation2_50: { count: val2_50, pct: (val2_50/totalExts*100).toFixed(2) },86            validation3_80: { count: val3_80, pct: (val3_80/totalExts*100).toFixed(2) },87            validation4_stability: { p0_5: stab_05, p1: stab_1, p2: stab_2, p5: stab_5 },88            validation5_outlier: {89                removedDl: validExtensions[0],90                new50Count: noOutlier_50,91                new80Count: noOutlier_80,92                newTop1Share: noOutlierTop1Share93            },94            optionA: {95                top1: (top1Sum / totalDownloads * 100).toFixed(2),96                next4: (next4Sum / totalDownloads * 100).toFixed(2),97                next15: (next15Sum / totalDownloads * 100).toFixed(2),98                bottom80: (bottom80Sum / totalDownloads * 100).toFixed(2)99            }100        }, null, 2));101    }102});103