Skip to content

meda_scplot

Module Group

src/eda1

Project Stage ID

4[^2]

Purpose

The purpose of this module is to provide the user with the ability to visualise each numerical columns in a pandas dataframe in a two dimensional figure relative to other numerical columns, the module revolves around the utilisation of seaborn

Location

Here are the locations of the relevant files associated with the module

module information:

/src/eda/meda_scplot.json

module activation functions:

/src/eda/meda_scplot.py

Requirements

Module import information

from mllibs.nlpi import nlpi
import seaborn as sns
import matplotlib.pyplot as plt
import pandas as pd
import math
from collections import OrderedDict
import warnings; warnings.filterwarnings('ignore')
from mllibs.nlpm import parse_json
import pkg_resources
import json

Selection

Activation functions need to be assigned a unique label. Here's the process of label & activation function selection 

def sel(self,args:dict):

    select = args['pred_task']

    if(select == 'col_kde'):
        self.eda_colplot_kde(args)
    elif(select == 'col_box'):
        self.eda_colplot_box(args)
    elif(select == 'col_scatter'):
        self.eda_colplot_scatter(args)

Activation Functions

Here you will find the relevant activation functions available in class meda_scplot

col_kde

description:

visualise/plot column feature kernel density estimation plot

code:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
def eda_colplot_kde(self,args:dict):

    # get numeric column names only
    num,_ = self.split_types(args['data'])

    if(args['x'] is not None):
        xloc = args['data'][args['x']]
    else:
        xloc = None

    if(args['hue'] is not None):
        hueloc = args['data'][args['hue']]
        if(type(nlpi.pp['stheme']) is str):
            palette = nlpi.pp['stheme']
        else:
            palette = palette_rgb[:len(hueloc.value_counts())]

    else:
        hueloc = None
        palette = palette_rgb

    columns = list(num.columns)  
    n_cols = 3
    n_rows = math.ceil(len(columns)/n_cols)

    fig, ax = plt.subplots(n_rows, n_cols, figsize=(16, n_rows*5))
    ax = ax.flatten()

    for i, column in enumerate(columns):
        plot_axes = [ax[i]]

        sns.set_style("whitegrid", {
        'grid.linestyle': '--'})

        sns.kdeplot(data=args['data'],
                    x=column,
                    hue=hueloc,
                    fill=nlpi.pp['fill'],
                    alpha= nlpi.pp['alpha'],
                    linewidth=nlpi.pp['mew'],
                    edgecolor=nlpi.pp['mec'],
                    ax=ax[i],
                    common_norm=False,
                    palette=palette
                     )

        # titles
        ax[i].set_title(f'{column} distribution');
        ax[i].set_xlabel(None)

    for i in range(i+1, len(ax)):
        ax[i].axis('off')

    plt.tight_layout()

col_box

description:

visualise/plot column feature boxplot

code:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
def eda_colplot_box(self,args:dict):

# split data into numeric & non numeric
num,cat = self.split_types(args['data'])

columns = list(num.columns)  
n_cols = 3
n_rows = math.ceil(len(columns)/n_cols)

if(args['x'] is not None):
    xloc = args['data'][args['x']]
else:
    xloc = None

if(args['x'] is not None):
    xloc = args['data'][args['x']]
else:
    xloc = None

if(args['hue'] is not None):
    hueloc = args['data'][args['hue']]
    if(type(nlpi.pp['stheme']) is str):
        palette = nlpi.pp['stheme']
    else:
        palette = palette_rgb[:len(hueloc.value_counts())]

else:
    hueloc = None
    palette = palette_rgb

fig, ax = plt.subplots(n_rows, n_cols, figsize=(16, n_rows*5))
sns.despine(fig, left=True, bottom=True)
ax = ax.flatten()

for i, column in enumerate(columns):
    plot_axes = [ax[i]]

    sns.set_style("whitegrid", {
    'grid.linestyle': '--'})


    if(args['bw'] is None):
        bw = 0.8
    else:
        bw = eval(args['bw'])

    sns.boxplot(
        y=args['data'][column],
        x=xloc,
        hue=hueloc,
        width=bw,
        ax=ax[i],
        palette=palette
    )

    # titles
    ax[i].set_title(f'{column} distribution');
    ax[i].set_xlabel(None)


for i in range(i+1, len(ax)):
    ax[i].axis('off')

plt.tight_layout()

col_scatter

description:

visualise/plot column feature scatterplot

code:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
    def eda_colplot_scatter(self,args:dict):

        # split data into numeric & non numeric
        num,_ = self.split_types(args['data'])

        columns = list(num.columns)  
        n_cols = 3
        n_rows = math.ceil(len(columns)/n_cols)

        if(args['x'] is not None):
            xloc = args['data'][args['x']]
        else:
            xloc = None

        if(args['hue'] is not None):
            hueloc = args['data'][args['hue']]
            if(type(nlpi.pp['stheme']) is str):
                palette = nlpi.pp['stheme']
            else:
                palette = palette_rgb[:len(hueloc.value_counts())]

        else:
            hueloc = None
            palette = palette_rgb

        fig, ax = plt.subplots(n_rows, n_cols, figsize=(16, n_rows*5))
        sns.despine(fig, left=True, bottom=True)
        ax = ax.flatten()

        for i, column in enumerate(columns):
            plot_axes = [ax[i]]

            sns.set_style("whitegrid", {
            'grid.linestyle': '--'})

            sns.scatterplot(
                y=args['data'][column],
                x=xloc,
                hue=hueloc,
                alpha= nlpi.pp['alpha'],
                linewidth=nlpi.pp['mew'],
                edgecolor=nlpi.pp['mec'],
                s = nlpi.pp['s'],
                ax=ax[i],
                palette=palette,
            )

            # titles
            ax[i].set_title(f'{column} distribution');
            ax[i].set_xlabel(None)


        for i in range(i+1, len(ax)):
            ax[i].axis('off')

        plt.tight_layout()
        plt.show()

  1. Reference to the sub folder in src ↩