SwiftKey Exploratory Data Analysis
Sameer Ahmad
2026-06-13
Introduction
The goal of this project is to develop a predictive text model capable of suggesting the next word based on a sequence of words entered by a user. The final application will be implemented using R and Shiny and will demonstrate basic natural language processing techniques.
The dataset used for this project consists of text collected from three sources:
- Blogs
- News articles
- Twitter posts
These datasets are provided in English and contain a large volume of real-world text suitable for language modeling.
Data Summary
The dataset contains text from:
| Source | Description |
|---|---|
| Blogs | Personal blog entries |
| News | News articles |
| Social media posts |
The combined dataset contains millions of words and thousands of lines of text.
Data Cleaning
Before analysis, the text data was cleaned by:
- Converting text to lowercase
- Removing punctuation
- Removing numbers
- Removing extra whitespace
- Removing special characters
These preprocessing steps help standardize the text for analysis.
Exploratory Analysis
The most common words found in the corpus include:
- the
- and
- to
- of
- a
These words appear frequently across all text sources.
Top Words
The figure below illustrates the most frequently occurring words in the dataset.
Bigram Analysis
Bigrams represent pairs of consecutive words.
Examples include:
- thank you
- i love
- good morning
- how are
- looking forward
Trigram Analysis
Trigrams represent sequences of three words.
Examples include:
- thank you for
- one of the
- i want to
- going to be
- looking forward to
Future Plans
The next stage of the project will focus on:
- Building n-gram language models
- Creating a predictive text algorithm
- Improving prediction accuracy
- Deploying the model using Shiny
- Publishing the application online
Conclusion
The exploratory analysis provided valuable insight into the structure of the dataset. Common words, bigrams, and trigrams were identified and will be used to construct a predictive text model for the final SwiftKey application.