[Python] How to make my model actually predict something? [closed]

I wanted to create a model that predicts age based on height, weight and sex of a person. For some reason it predicts always the same thing in test_model() function. In .csv file is everything fine I guess. I create random numbers and then in this code divide them by 1000 to make suitable for sigmoid function. I'm also not sure if I should use sigmoid. I was programming under the instructions of ChatGPT. Here is the code:

import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
from time import sleep

import matplotlib.pyplot as plt #pip install matplotlib
import numpy as np
import pandas as pd #pip install pandas

import tensorflow as tf #pip install tensorflow
from tensorflow.keras import layers
from tensorflow.keras.saving import register_keras_serializable
from tensorflow.keras.models import Model, load_model
from tensorflow.keras.callbacks import EarlyStopping

@register_keras_serializable()
class CustomModel(Model):
def __init__(self, **kwargs):
super(CustomModel, self).__init__(**kwargs)

self.dense1 = layers.Dense(64, activation='elu', kernel_regularizer=tf.keras.regularizers.l2(0.01))
self.dense2 = layers.Dense(32, activation='elu', kernel_regularizer=tf.keras.regularizers.l2(0.01))

self.output_layer = layers.Dense(1, activation='sigmoid')

def call(self, inputs):
x = self.dense1(inputs)
x = self.dense2(x)
return self.output_layer(x)

def get_config(self):
config = super(CustomModel, self).get_config()
return config


def train_model(self, x_train, y_train, batch_size, epochs):
self.compile(optimizer='adam',
loss='mse',
metrics=['mae'])

history = self.fit(x_train, y_train, batch_size=batch_size, epochs=epochs)
return history

def evaluate_model(self, x_test, y_test):
return self.evaluate(x_test, y_test)

def predict_data(self, x_test):
return self.predict(x_test)

def test_model(self):
print("Starting test", end="")
sleep(2)
print(".", end="")
sleep(1.5)
print(".", end="")
sleep(1.5)
print(".", end="\n\n")

while True:
prediction = 0
print("Write your age: ", end="")
age = int(input())
print("\nWrite your sex (0=Man, 1=Woman): ", end="")
sex = int(input())
print("\nWrite your height: ", end="")
height = int(input())
print("\nWrite your weight: ")
weight = int(input())

prediction = self.predict_data(np.array([[sex, weight, height]]))
print(f"Model thinks that your age is: {prediction[0][0]}")
print("Continue? y/n")
cont = input()
if cont.lower() == "y":
continue
else:
break


def main():
print("Train/Test?")
todo = input()
if todo.lower() == "train":
df_train = pd.read_csv("data/data_train.csv")
df_train["Weight"] = df_train["Weight"] / 1000
df_train["Height"] = df_train["Height"] / 1000
df_train["Age"] = df_train["Age"] / 1000
X_train = df_train[["Sex", "Weight", "Height"]].values
y_train = df_train["Age"].values
model = CustomModel()
history = model.train_model(X_train, y_train, epochs=2, batch_size=32)
print("=======================")
print("=======================")
df_test = pd.read_csv("data/data_test.csv")
df_test["Weight"] = df_test["Weight"] / 1000
df_test["Height"] = df_test["Height"] / 1000
df_test["Age"] = df_test["Age"] / 1000
X_test = df_test[["Sex", "Weight", "Height"]].values
y_test = df_test["Age"].values
predictions = model.predict_data(X_test)
test_loss, test_acc = model.evaluate_model(X_test, y_test)

print(f"Test Loss: {test_loss}")
print(f"Test Accuracy: {test_acc}")
model.save("model.keras")
model.test_model()


else:
model = load_model("model.keras")
model.test_model()




if __name__ == "__main__":
main()

print("Program finished")
input()

Continue reading...