def mlpGetPredictions(train, data, parameters, x_cols=DataColumns.getSelectedCols3()):
"""
Classify input data
Arguments:
train {array} -- labeled pandas dataframe
data {array} -- unlabeled pandas dataframe
parameters {namedTuple} -- parameters for classifier
"""
xtrain = train.loc[:, x_cols]
ytrain = train.loc[:, "label"]
xdata = data.loc[:, x_cols]
mlpClassifier = MLPClassifier(
hidden_layer_sizes=parameters.hidden_layer_sizes,
solver=parameters.solver,
alpha=parameters.alpha,
batch_size=parameters.batch_size,
learning_rate=parameters.learning_rate,
learning_rate_init=parameters.learning_rate_init,
max_iter=parameters.max_iter,
random_state=parameters.random_state,
verbose=parameters.verbose,
early_stopping=parameters.early_stopping,
validation_fraction=parameters.validation_fraction
)
mlpClassifier.fit(xtrain, ytrain.values.ravel())
ypred=mlpClassifier.predict(xdata)
return(ypred)
def getKnnPredictions(train, data, k, x_cols=DataColumns.getSelectedCols3()):
"""
Classify input data
Arguments:
train {array} -- labeled pandas dataframe
data {array} -- unlabeled pandas dataframe
k {int} -- number of nearest neighbors
Keyword Arguments:
x_cols {array} -- x column names (default: {DataColumns.getSelectedCols3()})
"""
xtrain = train.loc[:,x_cols]
ytrain = train.loc[:,"label"]
xdata = data.loc[:,x_cols]
knnClassifier = KNeighborsClassifier(n_neighbors=k, weights="uniform", metric="euclidean")
knnClassifier.fit(xtrain, ytrain.values.ravel())
ypred=knnClassifier.predict(xdata)
return(ypred)
data = DataHandler.calculateTotalForce(data)
data = DataHandler.calculateStepTime(data)
data = DataHandler.calculateForceValues(data)
data = DataHandler.calculateStepStartValues(data)
data = DataHandler.calculateStepMaxTimeValues(data)
data = DataHandler.calculateStepEndTimeValues(data)
data = DataHandler.calculatePhaseForceValues(data)
data = DataHandler.calculatePressTimeValues(data)
pd.set_option('display.max_columns', None)
data
#%% Bagging test
avg_acc, real_label, pred_label = Ensemble.testBagging(
data, DataColumns.getSelectedCols3())
pred_label_df = pred_label
real_label_df = real_label
pred_label_df = pred_label_df.replace("Normal", 0)
pred_label_df = pred_label_df.replace("Fall", 1)
real_label_df = real_label_df.replace("Normal", 0)
real_label_df = real_label_df.replace("Fall", 1)
avg_auc = roc_auc_score(real_label_df, pred_label_df)
print("AUC score: ", round(avg_auc, 2))
#%% 2d scatter
from sklearn.decomposition import PCA