import pandas as pd import numpy as np import matplotlib.pyplot as plt import seaborn as sns import statsmodels.api as sm import statsmodels.formula.api as smf %matplotlib inline

df = pd.read_csv("Schools.txt", sep=" ") df

df.round(2).head()

df.isna().sum()

df.describe()

fig, ax = plt.subplots(1, 4, figsize=(25, 7), constrained_layout = True) for variable, subplot in zip(df.columns, ax.flatten()): sns.boxplot(y = df[variable], ax = subplot)

import plotly.express as px import plotly.graph_objects as go from plotly.subplots import make_subplots variables = ["score", "class_size", "migration", "scholarship"] colors = ['#00afb9', '#FDD5AF', '#f07167', '#B185DB'] fig = make_subplots(rows=1, cols=4) for i, variable in enumerate(variables): trace = go.Box(y=df[variable], marker_color=colors[i], notched=True, name="") fig.add_trace( trace, row=1, col=i+1 ) fig.update_layout(height=500, width=1200, title_text="Box plot of each continuous variable", showlegend=False) fig.layout.template = 'plotly_white' fig.show()

df['migration_bin'] = pd.cut(df.migration,bins=[0,33,66,200],labels=['Low', 'Medium', 'High']) df['class_size_bin'] = pd.cut(df.migration,bins=[0,10,20,100],labels=['Low', 'Medium', 'High'])

df['migration_bin'].value_counts()

import plotly.figure_factory as ff group_labels=[ "small class", "medium class", "large class"] fig = ff.create_distplot( [ df[df["class_size_bin"] == "Low"]["score"], df[df["class_size_bin"] == "Medium"]["score"], df[df["class_size_bin"] == "High"]["score"] ], group_labels=group_labels, show_hist=False) fig.layout.template = 'plotly_white' fig.show()

import plotly.express as px import plotly.graph_objects as go from plotly.subplots import make_subplots group_labels=[ "small class", "medium class", "large class"] bins = ["Low", "Medium", "High"] fig = make_subplots(rows=3, cols=1) for i, mybin in enumerate(bins): trace = go.Box(x=df[df['class_size_bin'] == mybin]['score'], notched=True, name="") fig.add_trace( trace, row=i+1, col=1) fig.update_layout(height=350, width=1200, showlegend=False) fig.layout.template = 'plotly_white' fig.show()

df.head()

import plotly.express as px

fig = px.histogram(df["score"]) fig.show()

fig = px.histogram(df["scholarship"]) fig.show()

fig = px.histogram(df["class_size"]) fig.show()

fig = px.histogram(df["migration"]) fig.show()

df_corr = df.corr() print(df_corr)

%config InlineBackend.figure_format='retina' fig, ax = plt.subplots(figsize = (7, 5)) sns.heatmap(df_corr, annot = True, cmap="Blues")

import plotly.express as px px.imshow(df_corr, color_continuous_scale="blues")

# Lets plot a pairplot %matplotlib inline sns.set_context('paper') pairplot = sns.PairGrid(df) pairplot.map_upper(sns.scatterplot, linewidths = 1, edgecolor = 'w', s = 10) pairplot.map_diag(plt.hist) pairplot.map_lower(sns.kdeplot)

model1 = smf.ols("score ~ class_size + class_size + scholarship + migration", data=df).fit()

print(model1.summary())

model2 = smf.ols("score ~ class_size + I(class_size**2)+ scholarship + migration", data=df).fit() print(model2.summary())

fig = px.bar(model1.params[1:], orientation="h", text=round(model1.params[1:], 3) ) fig.layout.template = 'plotly_white' fig.show()

%config InlineBackend.figure_format='retina' # Coefficient confidence interval centers = model1.params[1:] conf_ints = model1.conf_int().iloc[1:, :] sort_index = np.argsort(centers.values) centers = centers[sort_index] conf_ints = conf_ints.iloc[sort_index] plt.barh(y = range(len(centers)), left = conf_ints[0], width = conf_ints[1] - conf_ints[0], height = 0.2, color= "#1460A8") plt.yticks(range(len(centers)), conf_ints.index) plt.plot(centers, range(len(centers)), 'ro') plt.plot([0, 0], [-0.5, len(centers) - 0.5], color = 'gray', linestyle = 'dashed') plt.xlim((-2, 2)) plt.ylim((-0.5, len(centers) - 0.5)) plt.title('Confidence interval explanatory variables')

residuals = model1.resid residuals

fig = px.scatter(x=model1.fittedvalues, y=residuals) fig.show()

fig = px.histogram(residuals, title="Distribution of the residuals") fig.layout.template = 'plotly_white' fig.show()