Sometimes Deep Sometimes Learning

Experimenting with Machine Learning and Deep Learning in Python.

I mostly use Jupyter Notebook, with Anaconda distribution of Python when it comes to DS/ML/DL.

Kaggle Profile : jaivarsan.

Analytics Vidhya Profile : greed2411.

Libraries used :

• scikit-learn : for purely Machine Learning Algorithms and preprocessing.
• pandas : for data analysis purposes.
• seaborn : for beautiful visualizations.
• plotly : for time series graphs.
• gensim : for NLP.
• keras : for deep learning purposes.
• tensorflow : for keras backend.
• pytorch : cause the hype is real.

Papers Implemented :

K. Altun, B. Barshan, and O. Tunçel, Comparative study on classifying human activities with miniature inertial and magnetic sensors, Pattern Recognition, 43(10):3605-3620, October 2010.

for UCI ML repositories : Daily and Sports Activities Data Set report and files available over here

Time for visualizations:

On Miscellaneous

Indian map plot of all the 498 cities mentioned in the NDL dataset from Kaggle:

Heat map from california house pricing dataset:

Object Detection using TensorFlow for both pictures and Video, for our Microprocessor Project using RPi:

Me and my friends, during a trek:

Friend of mine got caught on the RPi cam feed:

The famous iris dataset

• Andrews curves

• Parallel Plotting

• Pairplot

Data is Beautiful, Not Saturn's rings

On Healthcare

Breast Cancer

Melanoma Prediction

Cycling and Calories dataset

On Entertainment

The time when i solved math memes from facebook

In [28]: best_model_pred

Out[28]:

['7+2+5 = 107687\n',
 '7+2+5 = 191991\n',
 '7+2+5 = 202541\n',
 '7+2+5 = 18908\n',
 '7+2+5 = 183652\n']

On Business

Scatterplot on Relationship between Sales because of TV, Radio and Newspaper.

Plotly graphs of the recent trends in Bitcoin world, intereactive time series graphs.

Just to give you a tease:

This is the first dataset I handled, xD.

While working on NDL, A Kaggle dataset

I made the following observations:

Analysis made on the data set as of June 2017

• Number of cities : 493

• Number of states : 29

• Most number of cities are in the state: UTTAR PRADESH

• Number of cities in UTTAR PRADESH : 63

• Least number of cities belong to these states and their counts

  HIMACHAL PRADESH             1
  CHANDIGARH                   1
  TRIPURA                      1
  MIZORAM                      1
  NAGALAND                     1
  MANIPUR                      1
  MEGHALAYA                    1
  ANDAMAN & NICOBAR ISLANDS    1
  

• There are two Aurangabad(s) in the nation,

  • one belonging to BIHAR
  • second one belonging to MAHARASHTRA

• Top 5 States with the maximum number of Cities

    UTTAR PRADESH     63
    WEST BENGAL       61
    MAHARASHTRA       43
    ANDHRA PRADESH    42
    TAMIL NADU        32
  

• States vs City counts

  

• States vs District counts

  

• Each city and it's district number plot

  

  This graph made me analyse and conclude that

  • The most common district numbers are 11, 9 and 12 not the conventional 1, 2 and 3.

  • District Number, and their occurences and percentage they contribute to the total district count. Example : District Number 11, there are 37 districts in our India numbered 11, which contributes to 7.51% of total number of districts in India.

                                 District Counts  Percentage Index
     District Number                                   
     11                            37              7.51
     9                             26              5.27
     12                            24              4.87
     1                             22              4.46
     3                             22              4.46
     21                            21              4.26
    

  • 95.94% of districts have their district number value which is less than 50

• District numbers and their frequency

  

  • The above graphs tells us that there are no district numbers from 72 to 98 and few numbers here and there in the 40s & 50s

    Actual missing district numbers :

     40, 42, 43, 45, 51, 53, 55, 56, 58, 67, 69, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98
    

• Missing State Codes

   11, 12, 25, 26, 30, 31