Team 8

bonus_task_1.py

@@ -0,0 +1,46 @@
+#! /usr/bin/env python3
+import pandas as pd
+import numpy as np
+from sklearn import *
+
+## Training
+train_descriptors = pd.read_csv("../train_and_test_sets/train_descriptors.csv")
+train_mord3d = pd.read_csv("../train_and_test_sets/train_mord3d.csv")
+train_morgan = pd.read_csv("../train_and_test_sets/train_morgan.csv")
+train_rdk = pd.read_csv("../train_and_test_sets/train_rdk.csv")
+
+## Train responses
+train_crystals = pd.read_csv("../train_and_test_sets/train_crystals.csv")
+train_distances = pd.read_csv("../train_and_test_sets/train_distances.csv")
+train_centroid_distances = pd.read_csv("../train_and_test_sets/train_centroid_distances.csv")
+
+train_descriptors_full = train_descriptors.iloc[:, 3:-2].dropna(axis=1, how="any")
+train_mord3d_full = train_mord3d.dropna(axis=1, how="any").drop(['identifiers','Unnamed: 0', 'InchiKey', 
+                                                                 'smiles', 'name'], axis=1)
+train_morgan_full = train_morgan.drop('0',axis=1)
+train_rdk = train_rdk.drop('0',axis=1)
+
+## Testing
+test_descriptors = pd.read_csv("../train_and_test_sets/test_descriptors.csv")
+test_mord3d = pd.read_csv("../train_and_test_sets/test_mord3d.csv")
+test_morgan = pd.read_csv("../train_and_test_sets/test_morgan.csv")
+test_rdk = pd.read_csv("../train_and_test_sets/test_rdk.csv")
+
+test_descriptors_full = test_descriptors[train_descriptors_full.columns]
+test_mord3d_full = test_mord3d[train_mord3d_full.columns]
+test_morgan_full = test_morgan.drop('0',axis=1)
+test_rdk = test_rdk.drop('0',axis=1)
+
+train_PCA = decomposition.PCA(n_components=.95)
+scaler_for_PCA = preprocessing.StandardScaler()
+train_descriptors_PCA = train_PCA.fit_transform(scaler_for_PCA.fit_transform(train_descriptors_full))
+test_descriptors_PCA = train_PCA.transform(scaler_for_PCA.transform(test_descriptors_full))
+
+target = 'packing_coefficient'
+
+from sklearn import tree
+model = tree.DecisionTreeRegressor(criterion='mae')
+model.fit(train_descriptors_PCA, train_crystals[target])
+
+predictions = model.predict(test_descriptors_PCA)
+pd.DataFrame(predictions).to_csv("task_packing_coefficient_predictions22.csv", header=False, index=False)