BIT_BHAAI contribution to stocksight project

BIT_BHAAI_contri

@@ -0,0 +1,72 @@
+import pandas as pd
+import numpy as np
+import matplotlib.pyplot as plt
+import quandl
+
+data = quandl.get("NSE/TATAGLOBAL")
+
+plt.figure(figsize=(16,8))
+plt.plot(data['Close'],label='Closing Price')
+
+data['Open - Close'] = data['Open'] - data['Close']
+data['High - Low'] = data['High'] - data['Low']
+data=data.dropna()
+
+x=data[['Open - Close', 'High - Low']]
+
+#buy +1, sell -1
+#target variable y is for classification task
+y = np.where(data['Close'].shift(-1)>data['Close'],1,-1)
+
+from sklearn.model_selection import train_test_split as tts
+x_train, x_test, y_train, y_test = tts(x,y,test_size=0.25, random_state=44)
+
+#implementation of KNN classifier
+from sklearn.neighbors import KNeighborsClassifier 
+from sklearn import neighbors 
+from sklearn.model_selection import GridSearchCV 
+from sklearn.metrics import accuracy_score 
+
+
+#using gridsearch to find best parameter
+params = { 'n_neighbors': [2,3,4,5,6,7,8,9,10,11,12,13,14,15]}
+knn = neighbors.KNeighborsClassifier()
+model = GridSearchCV(knn, params, cv=5)
+
+#fit the model
+model.fit(x_train, y_train)
+
+#Accuracy score
+accuracy_train = accuracy_score(y_train, model.predict(x_train))
+accuracy_test = accuracy_score(y_test,model.predict(x_test))
+
+print('Train_data accuracy: ', accuracy_train)
+print('Test_data accuracy: ', accuracy_test)
+
+predictions_classification = model.predict(x_test)
+actual_predicted_data = pd.DataFrame({'Actual Class':y_test, 'Predicted Class': predictions_classification})
+
+y=data['Close']
+
+#implementation of knn regression
+from sklearn.neighbors import KNeighborsRegressor
+from sklearn import neighbors
+import sklearn
+x_train_reg, x_test_reg, y_train_reg, y_test_reg = sklearn.model_selection.train_test_split (x,y,test_size=0.25,random_state=44)
+
+#using gridsearch to find the best parameter
+params = {'n_neighbors': [2,3,4,5,6,7,8,9,10,11,12,13,14,15]}
+knn_reg=neighbors.KNeighborsRegressor()
+model_reg = GridSearchCV(knn_reg, params, cv=5)
+
+#fit the model and make predictions
+model_reg.fit(x_train_reg, y_train_reg)
+predictions = model_reg.predict(x_test_reg)
+print(predictions)
+
+#root mean square error
+rms=np.sqrt(np.mean(np.power((np.array(y_test)-np.array(predictions)), 2)))
+
+valid  = pd.DataFrame({'Actual Close': y_test_reg, 'Predicted Close Value': predictions})
+
+print(valid)