Sensor Data Fusion Binary Classification Algorithm
Budget: £20 – £250 GBP
I have data from 2 sensors (A and B). Both are used to measure the same set of samples. The sensors give a continuous response.
The samples are classified as true or false.
For example, consider one sample tested that is known to be true: sensor A gives a reading of 5682 and sensor B gives a reading of 2583.2
The data is attached in an excel sheet.
Outputs:
1) Can a single equation be given which is a function of the output from sensor A and the output from sensor B? The function should give a single value from an equation which is a combination of the result from sensor A and sensor B. The equation can be linear, non-linear, anything. The equation should be optimised. It should be optimised to maximise the performance of the binary classification.
2) From the single optimal equation that combines the out from sensor A and sensor B. What is the optimal threshold or cutoff to give the best classification.
Further considerations:
- Must be completed with Matlab. This can include any of the toolbox functions.
- Code should be fully commented and easy to understand.
- A short report should be provided. This will include: i) background/basics of the approach used, ii) explanation and figures to illustrate how the approach works, iii) figures and text to show that the approach is optimal, iv) figures and text to demonstrate the classification performance.
- Last point. It is possible that the approach outlined (above) is not the best. If you have an alternative that you can suggest, please do so.
The samples are classified as true or false.
For example, consider one sample tested that is known to be true: sensor A gives a reading of 5682 and sensor B gives a reading of 2583.2
The data is attached in an excel sheet.
Outputs:
1) Can a single equation be given which is a function of the output from sensor A and the output from sensor B? The function should give a single value from an equation which is a combination of the result from sensor A and sensor B. The equation can be linear, non-linear, anything. The equation should be optimised. It should be optimised to maximise the performance of the binary classification.
2) From the single optimal equation that combines the out from sensor A and sensor B. What is the optimal threshold or cutoff to give the best classification.
Further considerations:
- Must be completed with Matlab. This can include any of the toolbox functions.
- Code should be fully commented and easy to understand.
- A short report should be provided. This will include: i) background/basics of the approach used, ii) explanation and figures to illustrate how the approach works, iii) figures and text to show that the approach is optimal, iv) figures and text to demonstrate the classification performance.
- Last point. It is possible that the approach outlined (above) is not the best. If you have an alternative that you can suggest, please do so.