DSP Engineer turned Data Scientist @ccri
arrow = require('apache-arrow')
crimes = {
let response = await fetch(dataUrl);
let dataTable = await response.arrayBuffer().then(buffer => {
return arrow.Table.from(new Uint8Array(buffer));
});
return dataTable;
}
rowCount = crimes.count()
fields = crimes.schema.fields.map(f => f.name)
fieldTypes = crimes.schema.fields.map(f => f.type)
fieldTypeNames = crimes.schema.fields.map(f => f.type.toString())
firstRow = crimes.get(0).toString() // 1st row data
lastRow = crimes.get(rowCount-1).toString() // last row data
randomRow = crimes.get(Math.random() * rowCount)
toJSON = crimes.get(0).toJSON()
every10KRow = range(0, rowCount, 10000)
function range(start, end, step) {
let slice = [];
for (let i=start; i<end && i <rowCount; i+= step) {
slice.push(crimes.get(i).toArray());
}
return slice;
}
md`${getMarkdown(every10KRow)}`
function getMarkdown (dataFrame) {
let markdown = `${fields.join(' | ')}\n --- | --- | ---`; // header row
let i=0;
for (let row of dataFrame) {
markdown += '\n ';
let td = '';
let k = 0;
for (let cell of row) {
if ( Array.isArray(cell) ) {
td = '[' + cell.map((value) => value == null ? 'null' : value).join(', ') + ']';
} else if (fields[k] === 'Date') {
td = toDate(cell); // convert Apache arrow Timestamp to Date
} else {
td = cell.toString();
}
markdown += ` ${td} |`;
k++;
}
}
return markdown;
}
function toDate(timestamp) {
// Appache Arrow Timestamp is a 64-bit int of milliseconds since the epoch,
// represented as two 32-bit ints in JS to preserve precision.
// The fist number is the "low" int and the second number is the "high" int.
return new Date((timestamp[1] * Math.pow(2, 32) + timestamp[0])/1000);
}
latitude = columnStats('Latitude')
longitude = columnStats('Longitude')
function columnStats(columnName) {
const column = crimes.getColumn(columnName);
let max = column.get(0);
let min = max;
for (let value of column) {
if (value > max) {
max = value;
}
else if (value < min) {
min = value;
}
}
return {min, max, range: max-min};
}
function dateStats(columnName) {
const column = crimes.getColumn(columnName);
let max = toDate(column.get(0));
let min = max;
for (let value of column) {
let date = toDate(value);
if (date > max) {
max = date;
}
else if (date < min) {
min = date;
}
}
return {min, max, range: max-min};
}
timestampVector = crimes.getColumn('Date')
timestamps = {
let dates = []
let i = 0;
for (const timestamp of timestampVector.asEpochMilliseconds()) {
dates.push(new Date(timestamp));
if (++i >= 10) {
break;
}
}
return dates;
}
startDate = new Date(dates.max - 2*hour)
endDate = new Date(dates.max - 1*hour)
crimesIn1Hour = filterByDate(startDate, endDate)
function filterByDate(startDate, endDate) {
let lat, long, date, results = [];
const dateFilter = arrow.predicate.custom(i => {
let date = toDate(crimes.getColumn('Date').get(i));
return date >= startDate && date <= endDate;
}, b => 1);
crimes.filter(dateFilter)
.scan((index) => {
results.push({
'lat': lat(index),
'long': long(index),
'date': toDate(date(index))
});
}, (batch) => {
lat = arrow.predicate.col('Latitude').bind(batch);
long = arrow.predicate.col('Longitude').bind(batch);
date = arrow.predicate.col('Date').bind(batch);
}
);
return results;
}
days = Math.floor(dates.range / millisPerDay)
crimesInDays = filterByDate(startDateFilter, endDateFilter)
function scaleX(long) {
return (long - longitude.min) * (mapWidth/longitude.range);
}
function scaleY(lat) {
return mapHeight - (lat - latitude.min) * (mapHeight/latitude.range);
}
mappedCrimes = {
const canvas = DOM.canvas(mapWidth, mapHeight);
const context = canvas.getContext("2d");
context.fillStyle = '#333333';
let lat, long, x, y, count = 0;
const startDate = new Date(dates.min.getTime() + startMapDaySlider*millisPerDay);
const endDate = new Date(dates.min.getTime() + endMapDaySlider*millisPerDay);
const dateFilter = arrow.predicate.custom(i => {
let date = toDate(crimes.getColumn('Date').get(i));
return date >= startDate && date <= endDate;
}, b => 1);
crimes.filter(dateFilter)
.scan((index) => {
x = scaleX(long(index));
y = scaleY(lat(index));
context.fillRect(x, y, 1, 1); // rectWidth, rectHeight);
count++;
}, (batch) => {
lat = arrow.predicate.col('Latitude').bind(batch);
long = arrow.predicate.col('Longitude').bind(batch);
}
);
return {canvas, count};
}
One platform to build and deploy the best data apps
Experiment and prototype by building visualizations in live JavaScript notebooks. Collaborate with your team and decide which concepts to build out.
Use Observable Framework to build data apps locally. Use data loaders to build in any language or library, including Python, SQL, and R.
Seamlessly deploy to Observable. Test before you ship, use automatic deploy-on-commit, and ensure your projects are always up-to-date.