Titanic - Machine Learning from Disaster

This is my first Kaggle work where I have published my prediction and have improved my accuracy for the Titanic - Machine Learning from Disaster.

I had followed many notebooks and tutorials for creating this notebook.

The Work

    After a lot of studying and references to a lot of notebooks, I finally was able to understand this notebook. It is using a LogisticRegression model to obtain the prediction.
    The interesting things that I gathered from this notebook are that it had custom functions to get all the missing data and also many graphs to illustrate the missing data in both the training and the test datasets.
    Also, it is using a method to factorize the data for normalization and filling out the NaN's as -999 for Age and Fare. There was a detailed report taken with the use of the pandas_profiling library.
    With this as a base, I started my analysis.

    To improve the score from 0.77511 all I had to do was the following alterations:
  • Fill better mean values for the Age and Fare missing values instead of -999
  • Remove unwanted columns

Fill better mean values

    Filling mean median and mode values to NaN missing values will improve the model efficiency to a better amount in my case from 0.77511 to 0.77990.

    First I had taken the Correlation between the features as I had to group them on the basis of other correlated features so that I can get mean and fill in the missing values.

Filling Missing Values

Fare

Fig 1: Correlation Matrix between the features
    In my case, if you see Fig 1 it represents the correlation matrix between the different features. With this, I came to the conclusion that the Fare has a better relationship between  Parch and SibSp. This is very useful in filling missing values all we have to do is to group and aggregate the mean fare based on  Parch and we'll be able to improve our prediction.
Fig 2: Missing Values in the training dataset

Fig 3: Missing Values in the test dataset

Age

    For Age, I had the idea to extract the salutation from the name and group them together and aggregate the mean Age, and fillup the values based on that. Even then I found out that 'Ms' salutation was missing the age field in the test data. For this, I again ran through the same grouping aggregate through the Sex attribute.

Remove unwanted columns

Fig 4: Dropped Columns


    I never knew removing data had an impact on the model accuracy. After a lot of searching in the Kaggle forums, tutorials, and peer help (Swati Damele), I found that we have to remove unnecessary columns from the data that I had created for the Salutation. Removing this improved the model significantly from a score of 0.7790 to 0.78947

Future work to be done...

Fig 5: Box plot showing the outliers

    There are more things that can be done to improve the score like Removing outliers that I have found and recorded in the Age, SibSp, Parch, and Fare.

To Avoid

Fig 6: Removing outliers

    Must not use 25th quantile and 75th quantile to remove outliers. It decreased my accuracy from 0.78947 to 0.76315.

Conclusion

    Wrapping things up, the following were the observations:
  • Fill better mean values.
  • Remove unwanted columns.

Bibliography

Wholehearted thanks for helping me achieve this:
  • Swati Damele
  • https://www.kaggle.com/mishraay/titanic-dataset-prediction-easily-understandable
  • https://www.kaggle.com/mskbr1/titanic-project-kernel
  • https://www.kaggle.com/saisivasriram/titanic-feature-understanding-from-plots

Comments

Post a Comment

Popular posts from this blog

DISASTER ALERT SYSTEM

Image
  Introduction      From my Bachelor year, I have been seeing many of my close friends and relations dying due to road accidents. This is a serious problem which concerns me a lot. I decided from then on to follow this topic and the solutions that the major brains are figuring out.     This post is a solution for more border problems where we train a model for detecting disasters from tweets.     The project proposal can be found here . This is the first door that opens a wide project with numerous possibilities.      I used the existing tweets dataset from  Kaggle  to train my model and the  COLAB Notebook Dataset The following is the dataset Fig 1: Disaster Twiter feed Dataset The dataset consists of the id of the tweet, keywords in the tweet text, location of the tweet and the tweet text. The training set of the tweet alone has the target parameter which holds the resolution of the tweet. If the target is 0 th...
Read more

CIFAR10 Image Classifier using PyTorch

Image
     This is a blog post on the work that I have done in programming an image classifier for the CIFAR10 dataset.      I had followed many notebooks, tutorials, and guidance from my Teaching Assistant Mondol, Md Ashaduzzaman Rubel, and my fellow classmate Subbiah Sharavanan, Abishek Pichaipillai for creating  this notebook . The Work     First I had to work on a base tutorial code available on the PyTorch website here . It is a beginner tutorial for classifying images into their respective labels which are present in the CIFAR10 dataset. Also need to try different optimizers provided by PyTorch and document the results and infer some conclusions.      CIFAR10 dataset:      The CIFAR10 dataset consists of 6000 images with which there are 1000 tests and 5000 training images with labels.     It has images from 10 classes namely  plane, car, bird, cat, deer, dog, frog, horse, ship, and truck. The ...
Read more

Sentiment Prediction using Naive Bayes Algorithm

Image
Introduction This is a post about Sentiment Prediction work that I did with the Naive Bayes Classifier. The dataset I used for the experiments were on  this sentiment labelled dataset . Which had 3 types of review datasets: IMDB Movie Reviews Amazon Product Reviews Yelp Reviews I used the IMDB Movie Reviews dataset whose textual variation can be found here . The Jupyter Notebook having the outcome of my experimentation is committed in this  GitHub Repository . Dataset The dataset consists of review and sentiment pairs as follows Figure 1: The IMDB review dataset The reviews consist of movie reviews having both positive and negative sentiments. Each review is labelled with the respective sentiments which have positive as 0 and negative as 1. Goals The major part of this project was to understand the working of the Naive Bayes Classifier. The following are the important MVPs of the project: Predicting the sentiment for a given review. Dividing the dataset into train, dev and tes...
Read more