機械学習コンテストサンプルソース

estate.c

#include <float.h>
#include <limits.h>
#include <math.h>
#include <stdio.h>

#define TRAIN_SET 500
#define TEST_SET 500
#define FEATURE 6

#define K 3
#define SA 10000
#define FOLD 2
#define RANGE 10

int train[TRAIN_SET][FEATURE], label[TRAIN_SET], test[TEST_SET][FEATURE], result[TEST_SET];

void knn(int train_set[][FEATURE], int train_label[], int train_size, int test_set[][FEATURE], int test_label[], int test_size, int weight[])
{
    int i, j, k, sum, count, neighbors[K];
    double distance, dists[K];

    /*
    for (i = 0; i < FEATURE; i++)
        printf("%d ", weight[i]);
    printf("¥n");
    */

    for (i = 0; i < test_size; i++) {
        for (j = 0; j < K; j++)
            dists[j] = DBL_MAX;

        for (j = 0; j < train_size; j++) {
            distance = 0;
            for (k = 0; k < FEATURE; k++)
                distance += pow(train_set[j][k] - test_set[i][k], 2) * weight[k];
            distance = sqrt(distance);

            for (k = 0; k < K; k++) {
                if (distance < dists[k]) {
                    dists[k] = distance;
                    neighbors[k] = train_label[j];
                }
            }
        }

        sum = 0, count = 0;
        for (j = 0; j < K; j++) {
            if (dists[j] != DBL_MAX) {
                sum += neighbors[j];
                count += 1;
            }
        }

        test_label[i] = (int)round((double)sum / count);
    }
}

int main(void)
{
    int i, j, k, idx, range, old, diff, sum, weight[FEATURE] = {1};

    freopen("train.csv", "r", stdin);

    for (i = 0; i < TRAIN_SET; i++) {
        for (j = 0; j < FEATURE; j++)
            scanf("%d,", &train[i][j]);
        scanf("%d", &label[i]);
    }

    knn(train, label, TRAIN_SET / FOLD, &train[TRAIN_SET / FOLD], result, TRAIN_SET - TRAIN_SET / FOLD, weight);
    
    diff = 0;
    for (i = 0; i < TRAIN_SET - TRAIN_SET / FOLD; i++)
        diff += abs(label[TRAIN_SET / FOLD + i] - result[i]);

    srand(time(NULL));

    for (i = 0; i < SA; i++) {
        idx = rand() % FEATURE;
        range = rand() % RANGE;
        
        old = weight[idx];
        if (weight[idx] < range || rand() % 2 == 0)
            weight[idx] += range;
        else
            weight[idx] -= range;

        knn(train, label, TRAIN_SET / FOLD, &train[TRAIN_SET / FOLD], result, TRAIN_SET - TRAIN_SET / FOLD, weight);

        sum = 0;
        for (j = 0; j < TRAIN_SET - TRAIN_SET / FOLD; j++)
            sum += abs(label[TRAIN_SET / FOLD + j] - result[j]);

        if (sum > diff)
            weight[idx] = old;
    }

    freopen("test.csv", "r", stdin);

    for (i = 0; i < TEST_SET; i++) {
        for (j = 0; j < FEATURE - 1; j++)
            scanf("%d,", &test[i][j]);
        scanf("%d", &test[i][j]);
    }

    knn(train, label, TRAIN_SET, test, result, TEST_SET, weight);

    for (i = 0; i < TEST_SET; i++)
        printf("%d¥n", result[i]);

    return 0;
}

mushroom.c

#include <float.h>
#include <limits.h>
#include <math.h>
#include <stdio.h>

#define TRAIN_SET 1000
#define TEST_SET 1000
#define FEATURE 19
#define K 3

int train[TRAIN_SET][FEATURE], label[TRAIN_SET];

int main(void)
{
    int i, j, k, sum, distance;
    int test[FEATURE], dists[K], neighbors[K];

    freopen("train.csv", "r", stdin);

    for (i = 0; i < TRAIN_SET; i++) {
        for (j = 0; j < FEATURE; j++)
            scanf("%d,", &train[i][j]);
        scanf("%d", &label[i]);
    }

    freopen("test.csv", "r", stdin);

    for (i = 0; i < TEST_SET; i++) {
        for (j = 0; j < FEATURE - 1; j++)
            scanf("%d,", &test[j]);
        scanf("%d", &test[j]);

        for (j = 0; j < K; j++)
            dists[j] = INT_MAX;

        for (j = 0; j < TRAIN_SET; j++) {
            distance = 0;
            for (k = 0; k < FEATURE; k++)
                if (train[j][k] != test[k])
                    distance += 1;

            for (k = 0; k < K; k++) {
                if (distance < dists[k]) {
                    dists[k] = distance;
                    neighbors[k] = label[j];
                }
            }
        }

        sum = 0;
        for (j = 0; j < K; j++)
            sum += neighbors[j];
        printf("%d¥n", (int)round((double)sum / K));
    }

    return 0;
}