PCA_python_BiPlot_sample_Code.qmd

---
title: "PCA With Python"
author: "Habib Ezatabadi"
format: gfm
editor: visual
---


## import require libraries 

```{python}
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
```


## create a pca model 

```{python}
from sklearn.preprocessing import StandardScaler
from sklearn.decomposition import PCA
import plyer

## get path
pathh = plyer.filechooser.open_file()[0]

```


## implement Model 


```{python}
dat = pd.read_excel(pathh)
dat.shape
index = dat['RTO_Name'].values
ind = [2, 3, 8, 9, 10, 11, 12, 13, 14, 15, 16]
dat2 = pd.DataFrame(dat.iloc[:, ind])
dat2.index = index
dat2

## To define a standardizer for the data
scaler = StandardScaler() 
```





```{python}
colNames = dat2.columns
Index = dat2.index
dat_scaled = pd.DataFrame(scaler.fit_transform(dat2), columns = colNames, 
index = Index)
dat_scaled

```




## Obtaining all the main components

```{python}
from sklearn.preprocessing import MinMaxScaler
scaler2 = MinMaxScaler(feature_range=(0, 1))
pca = PCA(0.8) 
pca.fit(dat_scaled)
pca.n_components_  ## The number of components that cover 90% of the variance

Loading = pca.components_

temp1 = np.repeat("PC", 4); temp2 = [1, 2, 3, 4]
temp3 = list(map(lambda x, y: x + str(y), temp1, temp2))
df_loading = pd.DataFrame(Loading.T, index = colNames, columns = temp3)
df_loading


```

## create biplot


```{python}
from adjustText import adjust_text
#| fig-width: 9
#| fig-height: 9
Scores = pca.transform(dat_scaled)

df_score = pd.DataFrame(scaler2.fit_transform(Scores[:, 0:2]), 
index = index, columns = ["PC1", "PC2"])



def abbreviate(strings, length = 4):
    if len(strings) > length:
        ri = round(length/2)
        le = length - ri
        res = strings[:le] + strings[(len(strings) - le):len(strings)]
    else:
        res = strings
    return res

fig, ax = plt.subplots(figsize=(14, 9))
for i, feature in enumerate(df_loading.index):
    ax.arrow(0, 0, df_loading.iloc[i, 0], 
    df_loading.iloc[i, 1])



Texts = [ax.text(df_loading.iloc[i, 0] * 1.01, 
        df_loading.iloc[i, 1] * 1.01, 
        abbreviate(feature), fontsize=18, color = "purple", 
        style = "italic") for i, feature in enumerate(df_loading.index)]
adjust_text(Texts)

ax.scatter(df_score.PC1, df_score.PC2)
ax.set_xlabel('PC1', fontsize=20)
ax.set_ylabel('PC2', fontsize=20)
ax.set_title('Figure 1', fontsize=20)
plt.show()
```