
import tensorflow as tf
import matplotlib.pyplot as plt
import pandas as pd
import time
train = pd.read_csv("mnist_train(1).csv")
test = pd.read_csv("mnist_test(1).csv")
x_train = train.iloc[:,1:].values.astype('float32') / 255.0
y_train = train.iloc[:,0].values
x_test = test.iloc[:,1:].values.astype('float32') / 255.0
y_test = test.iloc[:,0].values
print("Training:", x_train.shape)
print("Testing:", x_test.shape)
experiments = [
    ("exp1",0.001,128,32,1),
    ("exp2",0.0001,128,32,1),
    ("exp3",0.01,128,32,1),
    ("exp4",0.001,64,32,1),
    ("exp5",0.001,128,64,1),
    ("exp6",0.001,128,32,2)
]
results = []
best_acc = 0
best_config = ""
for name,lr,batch,units,layers in experiments:
    model = tf.keras.Sequential([tf.keras.layers.Input((784,))])
    for i in range(layers):
        model.add(tf.keras.layers.Dense(units,activation='relu'))
    model.add(tf.keras.layers.Dense(10,activation='softmax'))
    model.compile(optimizer=tf.keras.optimizers.Adam(lr),loss='sparse_categorical_crossentropy',metrics=['accuracy'])
    log = tf.keras.callbacks.TensorBoard(log_dir=f"logs/{name}_{time.strftime('%H%M%S')}")
    start = time.time()
    h = model.fit(x_train,y_train,epochs=5,batch_size=batch,validation_split=0.1,callbacks=[log],verbose=0)
    test_acc = model.evaluate(x_test,y_test,verbose=0)[1]
    elapsed = time.time()-start
    results.append([name,lr,batch,units,layers,test_acc,elapsed])
    print(name,"Train:",round(h.history['accuracy'][-1],3),"Val:",round(h.history['val_accuracy'][-1],3),"Test:",round(test_acc,3),"Time:",round(elapsed,2))
    if h.history['val_accuracy'][-1] > best_acc:
        best_acc = h.history['val_accuracy'][-1]
        best_config = name
print("\nREPORT:")
print("Compared 6 MLP configurations.")
print("Parameters: Learning rate, batch size, units and layers.")
print("Best configuration:",best_config)
print("Best Validation Accuracy:",round(best_acc,3))
print("TensorBoard is used to visualize training performance.")
plt.bar([r[0] for r in results],[r[5]*100 for r in results])
plt.title("Accuracy Comparison")
plt.xlabel("Experiments")
plt.ylabel("Accuracy (%)")
plt.show()
plt.bar([r[0] for r in results],[r[6] for r in results])
plt.title("Training Time Comparison")
plt.xlabel("Experiments")
plt.ylabel("Time (seconds)")
plt.show()
print("\nRun: tensorboard --logdir=logs")
