| Revision | a84ddd6fd86cdf58bcf3c6889837eeee6b45dc5b (tree) |
|---|---|
| Zeit | 2015-03-20 06:36:08 |
| Autor | Lorenzo Isella <lorenzo.isella@gmai...> |
| Commiter | Lorenzo Isella |
A python code implementing a random forest for a Kaggle competition.
| @@ -0,0 +1,45 @@ | ||
| 1 | +#! /usr/bin/env python | |
| 2 | + | |
| 3 | +""" | |
| 4 | +Beating the benchmark | |
| 5 | +Otto Group product classification challenge @ Kaggle | |
| 6 | + | |
| 7 | +__author__ : Abhishek Thakur | |
| 8 | +""" | |
| 9 | + | |
| 10 | +import pandas as pd | |
| 11 | +import numpy as np | |
| 12 | +from sklearn import ensemble, feature_extraction, preprocessing | |
| 13 | + | |
| 14 | +# import data | |
| 15 | +train = pd.read_csv('train.csv') | |
| 16 | +test = pd.read_csv('test.csv') | |
| 17 | +sample = pd.read_csv('sampleSubmission.csv') | |
| 18 | + | |
| 19 | +# drop ids and get labels | |
| 20 | +labels = train.target.values | |
| 21 | +train = train.drop('id', axis=1) | |
| 22 | +train = train.drop('target', axis=1) | |
| 23 | +test = test.drop('id', axis=1) | |
| 24 | + | |
| 25 | +# transform counts to TFIDF features | |
| 26 | +tfidf = feature_extraction.text.TfidfTransformer() | |
| 27 | +train = tfidf.fit_transform(train).toarray() | |
| 28 | +test = tfidf.transform(test).toarray() | |
| 29 | + | |
| 30 | +# encode labels | |
| 31 | +lbl_enc = preprocessing.LabelEncoder() | |
| 32 | +labels = lbl_enc.fit_transform(labels) | |
| 33 | + | |
| 34 | +# train a random forest classifier | |
| 35 | +clf = ensemble.RandomForestClassifier(n_jobs=-1, n_estimators=500) | |
| 36 | +clf.fit(train, labels) | |
| 37 | + | |
| 38 | +# predict on test set | |
| 39 | +preds = clf.predict_proba(test) | |
| 40 | + | |
| 41 | +# create submission file | |
| 42 | +preds = pd.DataFrame(preds, index=sample.id.values, columns=sample.columns[1:]) | |
| 43 | +preds.to_csv('benchmark.csv', index_label='id') | |
| 44 | + | |
| 45 | +print "So far so good" |
Fork
Python-codes/otto-benchmark.py