# in this example, we'll create a heatmap of our correlation matrix
plt.figure(figsize=(16,16))
sns.heatmap(df.corr(), annot=True, fmt=".2f")

print(df.corr()['price_range'].sort_values(ascending=False))

# X = df[['ram','battery_power','px_width', 'px_height', 'int_memory', 'sc_w', 'sc_h', 'pc', 'talk_time', 'mobile_wt']]


# trying with a new phone

# new phone, outside the dataset
# for exmaple, if we we used fields: ['ram','battery_power','px_width', 'px_height']

# then, do something like this:
# X = df[['ram','battery_power','px_width', 'px_height', 'int_memory', 'sc_w', 'sc_h', 'pc', 'talk_time', 'mobile_wt']]
np_ram = 4096
np_battery = 1500
np_px_width = 600
np_px_height = 1200
np_intmemory = 32
np_scw = 5
np_sch = 12
np_pc = 24
np_talk_time = 12
np_mobile_wt = 150

# build the phone
new_phone = [[np_ram, np_battery, np_px_width, np_px_height, np_intmemory, np_scw, np_sch, np_pc, np_talk_time, np_mobile_wt]]

# what's the price range? scale the data and give it to model
new_phone = sc.transform(new_phone)
result = categories[np.argmax(model.predict(new_phone))]
print(f"\nPredicted price category: {result}")