twitter-graph : auto-report

WARNING: This repo is under test !

The main objective is to write an automatic report to facilitate the analysis of a twitter graph.

Twitter-graph : To know how to make a twitter-graph, look eleurent's repo:

Installation

Testing with python3.8

Need pip packages:

pip packages	tested version
python-docx	0.8.11
numpy	1.19.5
pandas	0.25.3
igraph	0.9.9
textdistance	4.2.2
matplotlib	3.4.2

Install requirements with :

pip3 install -r requirements.txt

How to use

from autoReport import *

autoReport("nodes.csv", "edges.csv", 
            {2339:"magenta", 
            2433:"limegreen",
            2566:"deepskyblue",
            1052:"saddlebrown",
            2498:"darkorange",
            1583:"red"}, 
           {"name":"Antoine Bondaz", 
           "at":"@AntoineBondaz",
            "description":"""Foodie - 🕵🏼‍Research @FRS_org - 👨🏼‍🏫 Teach @SciencesPo - 🇨🇳🇹🇼🇰🇷🇰🇵's foreign & security policies - Ph.D."""},
            rank_col_name="pageranks",
            class_col_name="modularity_class",
        ).makeReport("out.docx",
                      per_edeges=1,
                      show_internal_link=False)

Creates a report out.docx for Antoine Bondaz, with 6 class. nodes.csv have two columns : one with the rank of the nodes (default pageranks) and one with class (default modularity_class). If columns have special name, specify the argument class_col_name for class columns name and rank_col_name for rank column name in the construct.

Advice

- If your graph comes from the tweets search, you can add the tweets.json file with the tweets_path parameter (on the constructor) to have more graphics.
- To test quickly, you can add the per_edeges=0.1 parameter (to the makeReport method) to test the link graph with 10% of edges. Important to verify the correct installation of igraph

If the size of the nodes is calculated with In-Degree and In-Degree is not in the columns of the csv nodes, you can specify InDegree for the rank_col_name parameter, it will calculate it.

The report

The report is divided into 2 parts: the first one looks at the whole graph, the second one at each class.

Introduction

Class distribution

Links between class

Most common words

Similarity of descriptions

Class analysis

Class magenta ( 20.9%)

Most 10 common words