import pandas as pd import numpy as np import matplotlib.pyplot as plt import seaborn as sns from scipy.stats import skew from scipy.stats import probplot from scipy.stats import pearsonr

df_raw = pd.read_csv("../data/winequalityN.csv") df_raw.columns

df_raw.head()

df_raw.dtypes

df_raw.describe()

df_raw.isnull().sum()

df_raw = df_raw.interpolate()

df_raw.isnull().sum()

sns.set(rc={'figure.figsize': (14, 8)}) sns.countplot(x='quality', data=df_raw).set(title='Wine quality distribution') plt.show()

sns.pairplot(df_raw) plt.show()

sns.heatmap(df_raw.corr(), annot=True, fmt='.2f', linewidths=2) plt.show()

def plot_feature_distribution(df : pd.DataFrame(), feature : str) -> None: '''Plots probability and distribution analysis. ''' f, ax = plt.subplots(1, 2, figsize=(15, 5)) # Histogram sns.histplot(df[feature], kde=True, stat='density', linewidth=0, ax=ax[0]).set(title=f'{feature} distribution') #Probability plot probplot(df[feature], plot=ax[1]) plt.show() print(f'{feature} skewness = {skew(df[feature])}')

numeric_features = df_raw.select_dtypes(include=np.number).columns.tolist() for feature in numeric_features: plot_feature_distribution(df_raw, feature)

def get_correlation(feature_1 : str, feature_2 : str, df : pd.DataFrame()) -> None: pearson_corr, p_value = pearsonr(df[feature_1], df[feature_2]) print(f"Correlation between {feature_1} and {feature_2} is {pearson_corr}") print(f"P-value of this correlation is {p_value}")

sns.boxplot(x='quality', y = 'alcohol', data=df_raw, showfliers=False).set(title="Alcohol vs quality") plt.show() get_correlation('quality', 'alcohol', df_raw)

def convert_quality_numerical_to_categorical(value: int) -> None: if value <= 5: return 'low' elif value >5 and value <= 7: return 'medium' else: return 'high'

df_wine = df_raw df_wine['quality_label'] = df_wine['quality'].apply(convert_quality_numerical_to_categorical) #df_wine = df_wine.drop('quality', axis=1) df_wine.head()

print(df_wine['quality_label'].value_counts())

subset_attr = ['alcohol', 'density', 'volatile acidity', 'quality'] low = round(df_wine[df_wine['quality_label'] == 'low'][subset_attr].describe(), 2) medium = round(df_wine[df_wine['quality_label'] == 'medium'][subset_attr].describe(), 2) high = round(df_wine[df_wine['quality_label'] == 'high'][subset_attr].describe(), 2) pd.concat([low, medium, high], axis=1, keys=['Low Quality Wine', 'Medium Quality Wine','High Quality Wine'])

fig = df_wine.hist(bins=15, color='fuchsia', edgecolor='darkmagenta', linewidth=1.0, xlabelsize=10, ylabelsize=10, xrot=45, yrot=0, figsize=(10,9), grid=False) plt.tight_layout(rect=(0, 0, 1.5, 1.5)) plt.title('Numeric features distribution') plt.show()

fig, (ax) = plt.subplots(1, 1, figsize=(14, 8)) hm = sns.heatmap(df_wine.corr(), ax=ax, cmap="bwr", annot=True, fmt='.2f', linewidths=0.5) fig.subplots_adjust(top = 0.93) fig.suptitle('Wine Attributes and their Correlation Heatmap', fontsize=14, fontweight='bold') plt.show()

fig = plt.figure(figsize=(16, 8)) sns.countplot(data=df_wine, x="quality", hue="type").set(title="Type of wine frecuency distribution ") plt.show()

fig = plt.figure(figsize=(16, 12)) ax = fig.add_subplot(111, projection='3d') xscale = df_wine['residual sugar'] yscale = df_wine['free sulfur dioxide'] zscale = df_wine['total sulfur dioxide'] ax.scatter(xscale, yscale, zscale, s=50, alpha=0.6, edgecolors='w') ax.set_xlabel('Residual Sugar') ax.set_ylabel('free sulfur dioxide') ax.set_zlabel('Total sulfur dioxide') plt.title("Correlation between free/total sulfur dioxide and residual sugar") plt.show()

fig = plt.figure(figsize=(16, 12)) plt.scatter(x = df_wine['fixed acidity'], y = df_wine['free sulfur dioxide'], s = df_wine['total sulfur dioxide'] * 2, alpha=0.4, edgecolors='w') plt.xlabel('Fixed Acidity') plt.ylabel('free sulfur dioxide') plt.title('Wine free sulfur dioxide Content - Fixed Acidity - total sulfur dioxide', y=1.05) plt.show()