function featureSelector($wordbank_file)
{
global $wordBank, $positive, $negative, $total;
global $max_positive, $max_negative, $max_total;
readWordBankFromJSON($wordbank_file);
sortPositiveNegative();
$max_positive = max($positive);
$max_negative = max($negative);
$max_total = max($total);
print "\n" . $max_positive . "\t" . $max_negative . "\t" . $max_total . "\n";
print count($wordBank) . " => ";
removeLowFreqWords();
normalizePositiveLowFreqWords();
normalizeNegativeLowFreqWords();
writeWordBankToJSON($wordbank_file);
print count($wordBank) . "\n";
}
function countWordFreq($dataset_file, $wordbank_file = null)
{
if (!$wordbank_file) {
$wordbank_file = "w." . $dataset_file;
}
readWordBankFromJSON($wordbank_file);
$fp = fopen($dataset_file . ".csv", "r");
while (!feof($fp)) {
$tweet = fgetcsv($fp, 256);
if ($tweet) {
$tweet_words = featureExtractor($tweet[1]);
if ($tweet_words) {
updatewordBank($tweet_words, $tweet[0]);
}
}
}
writeWordBankToJSON($wordbank_file);
}
function featureSelector($wordbank_file)
{
global $wordBank, $pd, $total, $positive, $negative;
global $max_pd, $min_pd, $max_total, $max_positive, $max_negative;
readWordBankFromJSON($wordbank_file);
findProportionalDiff();
$max_pd = max($pd);
$min_pd = min($pd);
$max_total = max($total);
$max_positive = max($positive);
$max_negative = max($negative);
print $max_pd . " / " . $min_pd . "\t" . $max_total . "\n";
print count($wordBank) . " => ";
removeLowFreqWords();
removeLowCPDWords();
normalizePositiveLowFreqWords();
normalizeNegativeLowFreqWords();
writeWordBankToJSON($wordbank_file);
print count($wordBank) . "\n";
}
#!/usr/bin/env php
<?php
require_once "naive_bayes_classifier.php";
readWordBankFromJSON();
findEffeciency("testdata");
//sleep(18000);
function findEffeciency($file_name)
{
$correct = 0;
$incorrect = 0;
$fp = fopen($file_name . ".csv", "r");
while (!feof($fp)) {
$new_tweet = fgetcsv($fp);
if ($new_tweet) {
//print(findCatagory($new_tweet[1])."\n");
if (findCatagory($new_tweet[1]) == $new_tweet[0]) {
$correct++;
} else {
$incorrect++;
}
}
}
fclose($fp);
print "Correct:\t" . $correct . "\n";
print "Incorrect:\t" . $incorrect . "\n";
print "Effeciency:\t" . $correct / ($correct + $incorrect) . "\n";
}
function findCatagory($tweet)
{
$result = classify($tweet);
print_r($result);
