open in markdown viewer for more better view.

BID light sensor classifier algorithm challenge

Arduino Code in lighsensor_classifier Folder

Jotaro, 2020/01/10

Task

Your task is to complete int predict(int [], int) that predicts whether incoming data is black, gray or white. In this function you can use:

• array of obtained data so far, and its length

You must not edit any other lines in code except for above function.

Sensor data

Each data values belongs to one of three classes:{ 0-black, 1-gray, 2-white}. The data point follows three different Gaussian distribution with different mean and deviation unknown to each other.

Format

You don't have to code the data reading procedure but prepared code should do (tested with macOS Catalina clang environment).

N
d1 l1
d2 l2
...
dN lN

Evaluation

Your code will be tested both with existing test data (data1.txt - data4.txt) and will also be tested our private test data. The evaluation will be done by accuracy(%) (TODO: maybe I should use cross entropy).

Example

int predict(int* data, int num_prev_data){
    //your new data should be data[num_prev_data-1];
    int value = data[num_prev_data-1];

    //output: 0 for black, 1 for gray and 2 for white.
    if(value > 300)return 2;
    if(value > 200)return 1;
    else return 0;
}