Create probabilities for all chromosome frames

main.py

@@ -1,6 +1,8 @@
+import numpy as np
+
 from test_data_processor import get_test_data_frames, split_frames_by_is_valid
 from train_data_processor import prepare_features_labels, create_features
-from train_model import train
+from train_model import train_model
 
 alphabet = ["A", "C", "G", "T"]
 k = 4
@@ -9,14 +11,38 @@
 frame_length = 1500
 step = 750
 
+
+def get_probabilities_for_1_list(valid_frames, invalid_frames, y_test_proba):
+    prob_for_1 = [0] * (len(valid_frames) + len(invalid_frames))
+    print(len(prob_for_1))
+    for valid_frame_ind in range(len(valid_frames)):
+        prob_for_1_ind = int(valid_frames[valid_frame_ind].begin / step)
+        prob_for_1[prob_for_1_ind] = y_test_proba[valid_frame_ind][1]
+
+    mean_prob = np.mean(y_test_proba[:, 1], axis=0)
+
+    for invalid_frame_ind in range(len(invalid_frames)):
+        prob_for_1_ind: int = int(invalid_frames[invalid_frame_ind].begin / step)
+        prob_for_1[prob_for_1_ind] = mean_prob
+
+    return np.array(prob_for_1)
+
+
 if __name__ == "__main__":
     X_train, y_train = prepare_features_labels(train_filenames_labels, alphabet=alphabet, k=k)
-    clf = train(X_train, y_train)
+    clf = train_model(X_train, y_train)
 
     test_frames = get_test_data_frames(test_filename, frame_length=frame_length, step=step)
     valid_frames, invalid_frames = split_frames_by_is_valid(test_frames)
+
     X_test = create_features(valid_frames,
                              alphabet=alphabet,
                              k=k)
     y_test_proba = clf.predict_proba(X_test)
     print(y_test_proba)
+
+    print(len(valid_frames))
+    print(len(invalid_frames))
+    print("------------")
+    res = get_probabilities_for_1_list(valid_frames, invalid_frames, y_test_proba)
+    print(res)

test_data_processor.py

@@ -10,8 +10,8 @@ def get_test_data_frames(file_path, frame_length, step):
 
 def split_frames_by_is_valid(frames):
     return (
-        filter(lambda f: f.is_valid, frames),
-        filter(lambda f: not f.is_valid, frames)
+        list(filter(lambda f: f.is_valid, frames)),
+        list(filter(lambda f: not f.is_valid, frames))
     )
 
 

train_model.py

@@ -1,7 +1,7 @@
 from sklearn.ensemble import RandomForestClassifier
 
 
-def train(X, y):
+def train_model(X, y):
     clf = RandomForestClassifier(max_depth=5, random_state=123)
     clf.fit(X, y)
     return clf