gati’s gists

gati / guess_candidate_model.py

Created November 2, 2016 15:42

Code for training an LSTM model for text classification using the keras library (Theano backend). Was used for guesscandidate.com.

	from sklearn.cross_validation import train_test_split
	from keras.preprocessing import sequence, text
	from keras.models import Sequential
	from keras.layers import (Dense, Dropout, Activation, Embedding, LSTM,
	Convolution1D, MaxPooling1D)

	# Embedding
	max_features = 20000
	maxlen = 100
	embedding_size = 32

gati / largest_factor.py

Created November 17, 2014 01:59

Get the largest factor a number that's less than a threshold. Useful for finding the best number for breaking a list up into groups of equal size.

	import math

	"""
	Get the largest factor a number that's less than a threshold. Useful for finding the
	best number for breaking a list up into groups of equal size.
	"""
	def largest_factor(num, cur=1, factor=1, max_size=10):
	factor = math.ceil(float(num) / cur)
	if factor > max_size:
	return best_factor(arr, cur + 1, int(factor), max_size)

gati / where-ive-been.html

Created September 7, 2014 16:53

Visualize lat/lon points

	<!doctype html>
	<html>
	<head></head>
	<body>

	<!--
	Obviously this is not production code. Really I was just rendering a map so I could take a screenshot and post it on Twitter,
	and this was faster than using some Python mapping library (and looks way nicer!). So no judgement, mkay?

	This plots points taken from my Google Location data, formatted using this handy Python script:

gati / google_location.py

Created September 7, 2014 16:48

Format Google Location data

	import os
	import json
	import datetime

	"""
	I wanted to map where I'd been in the past year using DataMaps, a jQuery + D3
	library for visualizing geospatial data. https://github.com/markmarkoh/datamaps
	First I downloaded my data from Google https://www.google.com/settings/takeout
	(I let them track my Android phone), and then used the script below to
	format the JSON the way DataMaps expects.

gati / encoder.py

Created September 7, 2014 00:46

Mongoengine JSON encoder

	import json
	from collections import Iterable
	from bson import ObjectId
	from datetime import datetime

	class MongoengineEncoder(json.JSONEncoder):
	"""
	The default JSON encoder that ships with Mongoengine (the to_json method
	exposed on Document instances) makes some odd choices. Datetime objects
	are nested on a $date property, ObjectIds are nested on an $oid property,

gati / simple_closure.js

Created June 8, 2012 03:22

Simple JavaScript closure

	var takeMeTo = function(destination) {
	var refrain = 'Won\'t you take me to ' + destination;
	return function() {
	alert(refrain);
	}
	}

	// Nothing happens yet, you're just assigning the anonymous function returned
	// by takeMeTo to funkyTown
	var funkyTown = takeMeTo('Funky Town');

gati / dom_scraper.php

Created July 11, 2011 18:31

Quick scrape using simple_html_dom

	include('simple_html_dom.php'); // DOM parsing library.
	$url = (isset($_GET['site'])) ? $_GET['site'] : 'http://www.yelp.com'; //just an example, clean this
	$dom = file_get_html($url);
	foreach ($dom->find('a') as $node) {
	// Replace href attribute value
	$node->href = 'http://YOURPROXYSERVER.COM?requestedurl='.urlencode($url.$node->href);
	}
	// Output modified DOM
	echo $dom->outertext;

Jonathon Morgan gati