Update pages/4_Simple_EDA.py
Browse files- pages/4_Simple_EDA.py +101 -0
pages/4_Simple_EDA.py
CHANGED
|
@@ -0,0 +1,101 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
import pandas as pd
|
| 3 |
+
|
| 4 |
+
st.markdown("""
|
| 5 |
+
<style>
|
| 6 |
+
/* Set a soft background color */
|
| 7 |
+
body {
|
| 8 |
+
background-color: #eef2f7;
|
| 9 |
+
}
|
| 10 |
+
/* Style for main title */
|
| 11 |
+
h1 {
|
| 12 |
+
color: black;
|
| 13 |
+
font-family: 'Roboto', sans-serif;
|
| 14 |
+
font-weight: 700;
|
| 15 |
+
text-align: center;
|
| 16 |
+
margin-bottom: 25px;
|
| 17 |
+
}
|
| 18 |
+
/* Style for headers */
|
| 19 |
+
h2 {
|
| 20 |
+
color: black;
|
| 21 |
+
font-family: 'Roboto', sans-serif;
|
| 22 |
+
font-weight: 600;
|
| 23 |
+
margin-top: 30px;
|
| 24 |
+
}
|
| 25 |
+
|
| 26 |
+
/* Style for subheaders */
|
| 27 |
+
h3 {
|
| 28 |
+
color: red;
|
| 29 |
+
font-family: 'Roboto', sans-serif;
|
| 30 |
+
font-weight: 500;
|
| 31 |
+
margin-top: 20px;
|
| 32 |
+
}
|
| 33 |
+
.custom-subheader {
|
| 34 |
+
color: black;
|
| 35 |
+
font-family: 'Roboto', sans-serif;
|
| 36 |
+
font-weight: 600;
|
| 37 |
+
margin-bottom: 15px;
|
| 38 |
+
}
|
| 39 |
+
/* Paragraph styling */
|
| 40 |
+
p {
|
| 41 |
+
font-family: 'Georgia', serif;
|
| 42 |
+
line-height: 1.8;
|
| 43 |
+
color: black;
|
| 44 |
+
margin-bottom: 20px;
|
| 45 |
+
}
|
| 46 |
+
/* List styling with checkmark bullets */
|
| 47 |
+
.icon-bullet {
|
| 48 |
+
list-style-type: none;
|
| 49 |
+
padding-left: 20px;
|
| 50 |
+
}
|
| 51 |
+
.icon-bullet li {
|
| 52 |
+
font-family: 'Georgia', serif;
|
| 53 |
+
font-size: 1.1em;
|
| 54 |
+
margin-bottom: 10px;
|
| 55 |
+
color: black;
|
| 56 |
+
}
|
| 57 |
+
.icon-bullet li::before {
|
| 58 |
+
content: "β";
|
| 59 |
+
padding-right: 10px;
|
| 60 |
+
color: black;
|
| 61 |
+
}
|
| 62 |
+
/* Sidebar styling */
|
| 63 |
+
.sidebar .sidebar-content {
|
| 64 |
+
background-color: #ffffff;
|
| 65 |
+
border-radius: 10px;
|
| 66 |
+
padding: 15px;
|
| 67 |
+
}
|
| 68 |
+
.sidebar h2 {
|
| 69 |
+
color: #495057;
|
| 70 |
+
}
|
| 71 |
+
/* Custom button style */
|
| 72 |
+
.streamlit-button {
|
| 73 |
+
background-color: #00FFFF;
|
| 74 |
+
color: #000000;
|
| 75 |
+
font-weight: bold;
|
| 76 |
+
}
|
| 77 |
+
</style>
|
| 78 |
+
""", unsafe_allow_html=True)
|
| 79 |
+
|
| 80 |
+
st.header("Simple EDAπ¬")
|
| 81 |
+
st.markdown('''
|
| 82 |
+
- Simple EDA is a part of life cycle in NLP where after collecting the raw data we need to perform simple eda which tells the quallty of the data
|
| 83 |
+
- Simpl EDA is not performed based on the probelm statement
|
| 84 |
+
- It checks the exploration of the data
|
| 85 |
+
''')
|
| 86 |
+
|
| 87 |
+
st.subheader("Major Simple EDAπ")
|
| 88 |
+
st.markdown('''
|
| 89 |
+
β Whether all the alphabets are in
|
| 90 |
+
- lower case
|
| 91 |
+
- upper case
|
| 92 |
+
- combination of lower and upper case
|
| 93 |
+
β Whether the collected text data contains any html / url tags
|
| 94 |
+
β Whether the collected text data contains any urls
|
| 95 |
+
β Whether the collected text data contains any mentions / hashtags
|
| 96 |
+
β Whether the collected text data contains any digits
|
| 97 |
+
β Whether the collected text data contains any punctuations
|
| 98 |
+
β Whether the collected text data contains any emojis
|
| 99 |
+
β Whether the collected text data contains any data /time
|
| 100 |
+
''')
|
| 101 |
+
|