ce.tech AIaaS Case Studies
Full Stack Data Science Solutions
Supermarket sweep
A look at twitter sentiment around five major supermarket chains over the evening peak on Wednesday 22nd Feb 2023.
The tweets are scraped from 16:00 to 21:00 within 1,000km of Manchester.
Here's what was found...
Trade support engine with live news tracker
Crypto trading engine built with Dash open-source code and hosted on Heroku.
This app also showcases a rolling news feed.
Have a go at placing some dummy trades using the Buy/Sell option
* please allow 30 seconds - 1 minute to load
Met Office Data Mining
Python API script to fetch & transform MetOffice 7-day forecast into a pandas dataframe (accessible via Binder) and Case Study for UK wind farm 15-min generation data sourcing
Python script for scraping MetOffice 7-day forecast
Case Study: ce.tech ML forecast for UK Month Ahead energy prices
Case Study: ce.tech wind data sourcing and batch script API-extraction for UK wind farm generation at settlement period level from COD
Python script for scraping MetOffice 7-day forecast
* the GitHub python script here is a single module from a much larger UK electricity price forecast including UK load forecasts, wind and solar settlement period output, FX rates and gas prices as multiple macroeconomic and commercial drivers for wholesale month ahead power prices.
Deep Learning App with React & Flask
Image Classification for classification and archiving of 16 different document types.
After training a python deep learning model with TensorFlow and Colab GPU, we hooked it up to a Flask backend and react.js frontend and pushed the app to Heroku
You can play / test one of our apps using demo images here
* please allow 30 seconds - 1 minute to load
Dashboards, BI & Market Reporting
All businesses have to walk before they run and no AI project should start without a KPI overview anchored to the latest data - see the PowerBI, Dash (Python) and pdf reporting via excel examples below, all data-pipeline engineered to hook up to the latest data.
Forecasting with RNNs and fbprophet
The use of AI in forecasting is trending right now. Whether its fbprophet, LinkedIn's Greykite, Unit8's Darts or via a recurrent neural network, python provides a host of new libraries to go beyond traditional naive, exponential smoothing or Arima-based approaches.
Tesla 100 day forecast using LSTMs
Google looking to bounce up following non-farm payrolls
ChatGPTs take on forecasting
Tesla 100 day forecast using LSTMs
Azure Cognitive Services
Microsoft Azure Cognitive Services is one of the better services out there for quickly performing natural language processing (NLP).
Here's a look at how Text Analytics works via a Python demo taking Trip Advisor reviews and using the Azure REST API to extract keywords and sentiment after an initial OCR process with pytesseract...
AI app-hosting with AWS Lambda
One of the most popular serverless components on cloud is AWS Lambda.
The attraction of serverless computing is in focussing on app development without having to worry about servers.
AWS Lambda is based on Lambda architecture where Lambda functions (Function-as-a-Service) can facilitate the storage and processing of data ingested in both batches and in real-time (i.e. streaming).
Lambda functions can also serve web pages - take a look at Part I of our video series on deploying a react app with AWS Lambda and watch out for Part II coming soon on consuming a trained deep learning model (hosted on S3) via an enhanced GUI.
See our GitHub repo - 39 hands-on labs underpinning "Productionizing AI with Cloud & Python" (Published by Apress / Springer Nature)
or get in touch for more information below
