Using Big Data for a cattle-breeding concern
IT Svit had a customer from the United Kingdom, a stock-breeding concern specializing at breeding pigs. The customer had a centralized reporting system in place, which allowed the farmers to report the farm operational parameters like:
- The quantity of livestock,
- The numbers of animals gotten ill,
- Symptoms of the disease
- Types of medication administered
- How many animals recuperated or perished after treatment
Project requirements
The customer wanted us to provide the following services:
- Development of a Big Data solution for processing the data inputted by the farmers and delivering actionable business insights based on it
- Improvement of the efficiency of treatment by decreasing the time needed by the vets to assess the situation when the stock becomes ill
- A configurable and simple dashboard enabling the users to visualize the data in the form of linear graphs and compare multiple graphs with each other.
Project results
After we trained the Machine Learning (ML) algorithm based on the historical data, the system delivered the following benefits:
- Diagnosys of the disease based on the symptoms,
- Recommendations for treatment at once,
- Prescriptive analytics helped the customer to reduce the time between identifying the disease and applying the treatment,
- Significantly higher percentages of recuperated livestock (approx 40% more)
- Increased business profitability.
Location: The United Kingdom
Partnership period: September – December 2017
Team size: 2 people
Team location: Kharkiv, Ukraine
Services: Cloud infrastructure design and development, Big Data solutions, Web Development, Python development, data visualization, data normalization
Expertise delivered: AWS cloud administration, Machine Learning, DevOps services, Big Data architecture, Python development, data science, data visualization
Technologies: PostgreSQL, ML in AWS, Python, Flask, Jupyter Notebook, Chart.js
Product Overview
Client’s goals
The main goal was to increase the business profitability and reduce the losses due to livestock diseases. It was composed of several parts:
- Normalization of the existing data set (some users wrote the medication titles in capital letters and some in plain text, etc.)
- Creating and training a Machine Learning model to track the effectiveness of the meds
- Developing a dashboard for data visualization and turning the data analysis results into clearly understandable graphs
Implementation and challenges resolved
IT Svit team applied our experience with developing Big Data solutions and solved all the challenges:
- We wrote a data normalization tool in Pythonto ensure the uniformity of data
- We used Python and JupyteR Notebook to make statistical analysisand highlight patterns
- The underlying cloud infrastructure was deployed in AWS and we used the AWS Machine Learning service in the process of training the model
- The data was stored in the cloud using the PostgreSQL databaseand Amazon RDS
- We used Flask and Chart.js to enable the data visualization and created a flexible dashboard for viewing the graphs. The users can compare any pair of parameters as needed.