Prediction interval with examples of similar pattern and prediction strength

In this paper, we formed prediction intervals using historical similarities, found through the direct correlation. At first, a string of 5 to 20 recent samples is correlated with a long training string of samples. Then, the highest normalized correlation values and corresponding indexes are picked. After that, the amplitudes of the matched samples are adjusted by multiplying the value with the amplitude of recent string and by dividing by the amplitude of matched strings. These adjusted samples are actually the prediction values. Each prediction value is given a weight (relevance) based on the value of normalized correlation and a function of the ratio between amplitudes of strings. A bar chart is drawn using the weighted (relevance) distribution and less relevant regions are discarded from sides. A prediction strength is calculated from relevances. Except for the calculation of relevance, everything is calculated without any assumption. The user can check similar occurrences and decide to search more when the prediction strength is low.