Actable AI provides state-of-the-art level Causal AI and Machine Learning (ML)-based analytics, powered by open-source state-of-the-art Automated ML (AutoML) to train good models with little effort. It is not only a tool that helps to get prediction values, but also provides the rationale behind each prediction. Results generated by the model can either be downloaded or saved as a new table on Actable AI to be used for further analysis. Moreover, trained models are deployed instantly and can be used externally with a provided API, so that the model can be integrated into your existing applications (be it a web app, mobile app, etc.).
We cover the following ten machine learning-based analytics:
Data Cleanse & Imputation: Detect missing or invalid values and suggest fixed values to complete the data.
Causal Inference: Goes beyond simple associations between variables to estimate the causal relationships, and answer whether a change in a treatment variable affects an outcome.
Regression: A machine learning technique that predicts continuous values. This requires training data with ground truth values for each data point. An example use case is the prediction of rental prices of apartments.
Classification: A machine learning technique that classifies data points into different classes (or categories). This requires training data with class labels for each data point. Example use cases include predicting existing customers to be churn/not churn (allowing estimation of the churn rate) or predicting if a tumor is benign or malignant.
Segmentation: A machine learning technique that splits data points into segments, where data points in each segment are similar to each other. This is an unsupervised technique which does not require ground truth class labels as in the case of classification and regression.
Time-series Forecasting: Cutting-edge algorithm to predict future values of variables based on historical data. Examples include predicting the future price of a stock, the future price of a bond, the future price of a commodity, etc.
Counterfactual Analysis: Helps to predict outcomes when one of the inputs is intervened. This works by inferring the causal effect of the intervention using Machine Learning and generating predicted outcomes accordingly.
Sentiment Analysis: A Natural Language Processing (NLP) technique that extracts keywords and their sentiments from text input, to determine whether the sentiment of a piece of textual data is positive, negative, or neutral.
Optical Character Recognition (OCR): A technology that enables the conversion of printed or hand-written text (for example, as shown in an image) into machine-readable digital text.
Information Extraction: Process text to extract and organize any required information, using Large Language Models (LLMs) such as ChatGPT.
More details on the available functionalities may be viewed in their respective sections:
- Data Cleanse & Imputation
- Correlation Analysis
- Analysis of Variance
- Association Rules
- Bayesian Linear Regression
- Time-series Forecasting
- Causal Inference
- Causal Discovery
- Counterfactual Analysis
- Sentiment Analysis
- Optical Character Recognition (OCR)
- Information Extraction