import pandas as pd from sklearn.preprocessing import StandardScaler import seaborn as sns import numpy as np from tensorflow.keras import models, layers, regularizers

df = pd.read_csv("/content/sample_data/diamonds.csv")

df

df = pd.get_dummies(df, columns=["cut", "clarity", "color"])

X = df.columns.drop(["Unnamed: 0", "carat"]) #X = ["depth", "table", "price", "x", "y", "z"]

y = "carat"

scaler = StandardScaler() scaler = scaler.fit(df[X]) df[X] = scaler.transform(df[X])

sns.set(rc={"figure.figsize": (15,9)}) sns.distplot(df["carat"])

df = df[df[y] < np.mean(df[y]) + np.std(df[y]) * 3]

sns.distplot(df["carat"])

df_train = df[:45000] df_test = df[45000:]

model = models.Sequential() model.add(layers.Dense(32, activation="relu", input_shape=(26,), kernel_regularizer=regularizers.l2(0.001))) #model.add(layers.Dropout(0.5)) model.add(layers.Dense(32, activation="relu", kernel_regularizer=regularizers.l2(0.001))) #model.add(layers.Dropout(0.5)) model.add(layers.Dense(1))

model.compile(optimizer="rmsprop", loss="mse", metrics=["mae"])

history = model.fit(df_train[X], df_train[y], batch_size=512, epochs=50, validation_split=0.3)

sns.lineplot(x=range(0,50), y=history.history["val_loss"])

sns.lineplot(x=range(0,50), y=history.history["val_mae"])

model.evaluate(df_test[X], df_test[y])