Will a website redesign see an increased customer conversion? AB Testing explored.

In this project, I delve into AB Testing to see if a new website redesign will see an uplift in customer conversion. The company has run a randomised test, where visitors will be split in four groups, seeing

  • 1. The original website design and the original set of images
  • 2. The new website design and the new set of images
  • 3. The new website design but the original set of images
  • 4. The original website design but with the new set of images
  • I utilse both frequentism and bayesian methods to establish the probabity that a change will create an uplit in conversion, whether the uplift could be caused by random chance and what the mean uplift will likely be.

    Fake News? Hate Speech? Summarising news articles

    In this project, I delve into the world of natural language processing (NLP), where I create a class that will

  • 1. Scrape the text of any news article, given just a URL
  • 2. Warn whether that article may contain hate speech or fake news
  • 3. Provide a summary (a few sentences) of that article, no matter how long the article is.
  • I achieve this by creating two models, through the use of a single custom class and pipelines to predict if a given text may contain fake news or hate speech. I then take an extractive text summarisation approach to summarise the article.

    Emotional Speech Recognition

    Humans are really good at understanding the emotion behind spoken words. Does someone sound happy, or angry? Can we build an AI to do this and if so why would we want to when we're already so good at it?

    We're good at understanding emotion because it's vital to our communication. We react and respond differently to people depending on their emotional state, be it happy, sad, angry or another emotion.

    Having an AI complete this task can provide great advantage in so many different ways, many of which wouldn't be possible by us; from having digital assistance tools (think Siri and Alexa) understand the emotional intent of a request, to a business analysing their calls to identify potential complaints or the general mood of their customers. You could also apply this in a call center telephony routing, where the system picks up angry or upset customers and transfers them to a human to prevent further agitation.

    This is also where AI can really help because we can process audio in bulk or without human intervention. Rather than having a person listen to a sample of calls, an AI can listen to a large volume without an invasion of privacy, whilst outputting the general positive or negative nature of their customers.

    Analysis of GB's Road Traffic Incidents

    DataCamp's analysis of traffic incidents, asking
    How can we reduce fatalities?

    The Scenario: You work for the road safety team within the department of transport and are looking into how they can reduce the number of major incidents. The safety team classes major incidents as fatal accidents involving 3+ casualties. They are trying to learn more about the characteristics of these major incidents so they can brainstorm interventions that could lower the number of deaths. They have asked for your assistance with answering a number of questions.

    Marketing Promotions

    DataCamp's first competition asked the question
    Where should a drinks company run promotions?

    The Scenario: Your company owns a chain of stores across Russia that sell a variety of alcoholic drinks. The company recently ran a wine promotion in Saint Petersburg that was very successful. Due to the cost to the business, it isn’t possible to run the promotion in all regions. The marketing team would like to target 10 other regions that have similar buying habits to Saint Petersburg where they would expect the promotion to be similarly successful.

    Portfolio

    A selection of my public projects.

    AB Testing

    Can we establish if a new website design will increase customer conversion? Could this increase be due to random chance? What is the likely uplift we will see?

    Fake news? Hate News? NLP Summary

    Can we summarise news articles and warn users when it may (or may not) contain fake news or hate speech?

    Emotional Speech Recognition

    Can AI understand the emotional intent behind speech? What uses and benefits could this provide?

    House Prices v2

    Can we predict the value of a house? Detailed EDA of the Ames housing dataset.

    Road Traffic Incidents

    Using data from the UK department of transport, I look into how we can reduce the number of major incidents.

    Digit Recogniser

    Recognising hand writen numbers, through the use of PCA and SVC machine learning techniques on the MNIST ("Modified National Institute of Standards and Technology") dataset.

    Drinks Promotion

    Following a recent wine promotion, we use clustering techniques to identify 10 regions to target for our next wine marketing campaign.

    Used Car Prices

    DataCamp Professional Certificate in Data Science project.
    The sales team wants to know whether you can make predictions for the price of used cars.

    Births and Pregnancy in Scotland

    Through the University of Edinburgh, this was my submission for the 10-credit online course; Data Visualisation for Professionals.

    TV Ratings

    My girlfriend and I have recently started watching a few TV sitcoms that have recently finished. I thought it’ll be a good idea to visualise these shows and see how they stand up over their many seasons.

    House Prices

    Kaggle Competition.
    Can we predict the value of a house?