template for logistic regression w/ graph and all from csv file

lr_by_term.py

import numpy as np
from sklearn.cross_validation import train_test_split
from sklearn.linear_model import SGDClassifier
from sklearn.metrics import precision_recall_curve


dfp = PATH_TO_FILE_HERE # format is csv with itemid,label,features

def readff(fpath):
    trids, labs, feats = [], [], []
    for l in open(fpath):
        ll = l.strip().split(',')
        trids.append(ll[0])
        labs.append(int(ll[1]))
        feats.append(np.array([float(f) for f in ll[2:]]))
    return trids, np.array(labs), np.array(feats)

classifier = SGDClassifier(fit_intercept=False)

T, L, X = readff(dfp)

def shuffle(X, Y):
    s = np.random.get_state()
    np.random.shuffle(X)
    np.random.set_state(s)
    np.random.shuffle(Y)
    return X, Y



Tt, Tv, Lt, Lv, Xt, Xv = train_test_split(T, L, X, test_size=0.1)

Xt, Lt = shuffle(Xt, Lt)

classifier.fit(Xt, Lt)
Tv = classifier.decision_function(Xv)

clf()
pp, rr, _ = precision_recall_curve(Lv, Tv)
plot(rr, pp)
pp, rr, _ = precision_recall_curve(-Lv, 1 - Tv)
plot(rr, pp)