# Plot the data and curve fit
fig, ax = plt.subplots(1,1)
ax.plot(sorted_data, cprob, c='0.75', alpha=0.75, linestyle='-',
linewidth=2.)
ax.set_ylabel('Cumulative probability')
ax.set_ylim((0, 1.))
if data_label:
ax.set_xlabel(data_label)
return ax.figure
if __name__ == '__main__':
train = pd.read_csv('./data/train.csv', index_col='Id')
#test = pd.read_csv('./data/test.csv', index_col='Id')
train = encode_categorical(train)
#plot joint plots of each variable with Hazard
hazard_jointplot(train)
# Usage - to show only high hazard plots:
#threshold = math.exp(3)
#hazard_jointplot(train, './plots/high/', threshold=threshold)
# plot a histogram of the hazard
ax = sns.distplot(train.Hazard, fit=expon, kde=False)
ax.figure.savefig('./plots/hazard_hist.png')
#plot the cumulative distibution & fit exponential curve
g = plot_exp_fit(train.Hazard)
g.savefig('./plots/hazard_exp_fit.png')
import xgboost as xgb
from gini import Gini
from preprocess import encode_categorical
# function used to binarize the hazards
def binarizer(x, threshold):
return int(x > threshold)
train = pd.read_csv('./data/train.csv', index_col='Id')
test = pd.read_csv('./data/test.csv', index_col='Id')
columns = train.drop(['Hazard'], axis=1).columns
# encode categorical variables as numbers
train = encode_categorical(train)
test = encode_categorical(test)
code_id = 105
# train a random forest
n = 500
n_split = 2
low = np.arange(2,71)
#high = np.arange(35,75,5)
haz_bins = low #np.concatenate((low, high))
#weights = np.concatenate((np.tile([1], len(low)), np.tile([5], len(high))))
weights = np.tile([1], len(low))
haz_pred_i = np.zeros((test.shape[0], haz_bins.shape[0]))