Data analysis is a growth market. The details of human interactions, and knowledge can now be analysed for patterns of behaviour, through a variety of techniques. These analysis techniques can produce insights and predictions, which can guide the future actions of governments, and corporations.
Machine learning is a semi-autonomous activity performed by a computer system, with decreasing human input, which produces a prediction based on the available data.
It is a data analysis technique that uses the power of a computer to recognize patterns and predict outcomes. However, to reach the desired prediction, the machine must learn to refine the underlying algorithm, to remove errors and to produce accurate and repeatable results.
Training the computer system progresses in stages. Initially, data with a known outcome is provided for analysis. The computer, through pattern recognition and iterative processes, will adjust the algorithm until the known outcome is produced.
The second stage is to mix data with a known outcome, with unanalysed data. The computer needs to apply the algorithm to the mixed data and determine if the pattern and prediction is still true. The machine will continue to adapt the algorithm until a reliable and repeatable result is produced.
The final stage is for the machine to work with completely unanalysed data to produce a suitably reliable outcome. An outcome which is repeatable, and which predicts a future behaviour. Occasional analysis of the produced results by a data scientist, will help to minimise algorithmic bias.
Data science is a human activity, using data analysis techniques, and software manipulation, to produce insights.
This science is located at the intersection of statistics, social science, computer and information science, and design. Using experimental design, domain knowledge, statistical inference and date visualization techniques, a data scientist analyses the data, and communicates insights. The insights could be descriptive, exploratory or causal in nature.
As with machine learning, this technique is iterative, and needs to produce reliable results. A data scientist will use human interpretive skills, to adjust the analysis algorithm, to remove outlying data and algorithmic bias.
Data science involves an awareness of the consequences of the insights produced. The who, what, how and why of the data, is firmly in a data scientist’s mind, providing clarity and direction throughout the analysis.
The goal of artificial intelligence is to produce a machine which is intelligent and reacts like a human. But AI is a functional technique, before this goal is reached. Where a computer system executes and recommends actions autonomously, it could be considered an artificial intelligence. AI produces actions, not insights or predictions.
AI involves recognition of speech, how to learn, how to plan, and how-to problem solve. Artificial intelligence requires knowledge, reasoning, and perception, to understand and learn autonomously. It also needs the ability to move and manipulate objects.
The field of artificial intelligence is firmly intertwined with robotics, and at present, it draws heavily on the machine learning and data science techniques.