emlearn_iris

import os
import ipdb # ipdb.set_trace()
import numpy as np
from sklearn import datasets
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import OneHotEncoder, MinMaxScaler, StandardScaler
from sklearn.ensemble import RandomForestClassifier

from keras.models import Sequential
from keras.layers import Dense


data = datasets.load_iris()

X = data['data']
Y = data['target']

# use and the scaled input the encoded output
enc = OneHotEncoder()
Y_enc = enc.fit_transform(Y[:, np.newaxis]).toarray()
Y = Y_enc

scaler = MinMaxScaler()
X_scaled = scaler.fit_transform(X)
X = X_scaled

X_train, X_test, Y_train, Y_test = train_test_split(X, Y, test_size=0.3, random_state=0)



num_features = X.shape[1]
num_classes = Y.shape[1]

batch_size = 1
epochs = 100


model = Sequential()
model.add(Dense(5, input_dim=num_features, activation='tanh'))
model.add(Dense(num_classes, activation='softmax'))


model.summary()

model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
#model.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])

history = model.fit(X_train, Y_train,
                    batch_size=batch_size,
                    epochs=epochs,
                    verbose=1,
                    )

score_train = model.evaluate(X_train, Y_train, verbose=0)
print('Train loss:', score_train[0])
print('Train accuracy:', score_train[1])

score_test = model.evaluate(X_test, Y_test, verbose=0)
print('Test loss:', score_test[0])
print('Test accuracy:', score_test[1])



import emlearn
cmodel = emlearn.net.convert_keras(model, method='loadable') #methods = pymodule or loadable
cmodel.save(file='iris_keras.h', name='iris')