import matplotlib.pyplot as plt
from matplotlib.pyplot import annotate, savefig, rcParams, rc, tick_params, xlabel, ylabel, xlim, ylim, legend, title
from sklearn.metrics import roc_curve, auc
from sklearn.preprocessing import label_binarize
from .performance_functions import BinaryClassification
from sklearn.tree import export_graphviz
from subprocess import call
import os, shutil
import seaborn as sns
from itertools import combinations
import numpy as np
from pdpbox import pdp

def plot_pdp_plots(model,df,feature_list,ranked_features,ranked_labels):
    feature_pairs = np.array(list(combinations(ranked_features,2)))
    feature_labels = np.array(list(combinations(ranked_labels,2)))
    for i in range(len(feature_pairs)):
        inter1 = pdp.PDPInteract(model=model,df=df, model_features=feature_list,features=feature_pairs[i],feature_names=feature_labels[i], pred_func=None, n_classes=0, memory_limit=0.5, chunk_size=-1, n_jobs=1, predict_kwds=None, data_transformer=None, num_grid_points=[10, 10], grid_types='percentile', percentile_ranges=[(5, 95), (5, 95)], grid_ranges=None, cust_grid_points=None)        
        fig, axes = inter1.plot(plot_type='contour', plot_pdp=False, to_bins=True, show_percentile=False, which_classes=None, figsize=None, dpi=300, ncols=2, plot_params=None)
        fname = feature_pairs[i][0]+'_'+feature_pairs[i][1] +'.pdf'
        plt.tight_layout()
        plt.savefig(fname, dpi=500,format='pdf')