example shitty rf

rf_slugs.py

from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import confusion_matrix
from sklearn.cross_validation import train_test_split
import pandas as pd
import numpy as np

def training_to_XY(hi,side='buy'):
    X = pd.DataFrame(hi.implied_size)
    if side == "buy":
        X['Through2'] = hi.implied_trd-hi.AskPrice2
        X['Through5'] = hi.implied_trd-hi.AskPrice5
        X['Support2'] = hi.implied_trd-hi.BidPrice2
        X['Support5'] = hi.implied_trd-hi.BidPrice5
        X['bsize'] = hi.bidsize1
        X['asize'] = hi.asksize1
        X['face'] = deltas_in_face(hi.implied_trd,hi.ix[:,['AskPrice1','AskSize1','AskPrice2','AskSize2','AskPrice3','AskSize3','AskPrice4','AskSize4','AskPrice5','AskSize5',]].values,1)
        X['supporting'] = deltas_in_face(hi.implied_trd,hi.ix[:,['BidPrice1','BidSize1','BidPrice2','BidSize2','BidPrice3','BidSize3','BidPrice4','BidSize4','BidPrice5','BidSize5',]].values,-1)
        X['prev'] = hi.prev_tick
        X['bigsize'] = hi.implied_size > hi.asksize1
    else:
        X['Through2'] = hi.implied_trd-hi.BidPrice2
        X['Through5'] = hi.implied_trd-hi.BidPrice5
        X['Support2'] = hi.implied_trd-hi.AskPrice2
        X['Support5'] = hi.implied_trd-hi.AskPrice5
        X['bsize'] = hi.bidsize1
        X['asize'] = hi.asksize1
        X['face'] = deltas_in_face(hi.implied_trd,hi.ix[:,['BidPrice1','BidSize1','BidPrice2','BidSize2','BidPrice3','BidSize3','BidPrice4','BidSize4','BidPrice5','BidSize5',]].values,-1)
        X['supporting'] = deltas_in_face(hi.implied_trd,hi.ix[:,['AskPrice1','AskSize1','AskPrice2','AskSize2','AskPrice3','AskSize3','AskPrice4','AskSize4','AskPrice5','AskSize5',]].values,1)
        X['prev'] = hi.prev_tick
        X['bigsize'] = hi.implied_size > hi.asksize1
    Y = hi.tick
    return X,Y
    

def quick_eval(X,Y):
    X_train, X_test, Y_train, Y_test = train_test_split(X.values, Y.values, random_state=0,test_size=.4)
    forest = RandomForestClassifier(n_estimators = 100,compute_importances=True)
    Y_pred = forest.fit(X_train,Y_train).predict(X_test)
    cm = confusion_matrix(Y_test,Y_pred)
    print(cm)
    matshow(cm)
    title('Confusion matrix')
    colorbar()
    ylabel('True label')
    xlabel('Predicted label')
    show()
    print pd.DataFrame(zip(X.columns,forest.feature_importances_)).sort(columns=1)
    
training_store = pd.HDFStore('training_slugger.h5')
X,Y = training_to_XY(training_store['buys'],'buy')
quick_eval(X,Y)
X,Y = training_to_XY(training_store['sells'],'sell')
quick_eval(X,Y)
training_store.close()