OD DATA ANALYZING APPARATUS, OD DATA ANALYZING METHOD, AND PROGRAM

To obtain a base pattern of movement through analysis of OD data representing frequency of movement from an origin to a destination.SOLUTION: An OD data analysis apparatus 101 has an acquiring unit 102 for acquiring a frequency c'i,j of movement from a point i of N origins to a point j of M destinations. A calculating unit 103 calculates a probability matrix W' based on the acquired frequency c'i,j. A decomposing unit 104 subjects the probability matrix W', with respect to a base number K, to approximate decomposition to obtain a product TU^R^V^ of a transposed matrix TU^ of a non-negative matrix U^=[u^k,i] having K rows and N columns and having the sum of each row set as 1, a non-negative diagonal matrix R^=diag [r^k] having the sum of diagonal elements set as 1, and a non-negative matrix V^=[v^k,j] having K rows and M columns and having the sum of each row set as 1. An output unit 105 outputs r^k as a ratio of movement classified into a base pattern k, u^k,i as probability of departing from the point i, and v^k,j as probability of reaching the point j.SELECTED DRAWING: Figure 1 【課題】出発地から目的地までの移動の頻度を表すODデータを分析して、移動の基底パターンを得る。【解決手段】ODデータ分析装置101において、取得部102は、N個の出発地のうち地点iから、M個の目的地のうち地点jまでの移動の頻度c'i,jを取得する。計算部103は、取得された頻度c'i,jに基づいて、確率行列W'を計算する。分解部104は、基底数Kに対して、確率行列W'を、各行の総和が1であるK行N列の非負行列U^ = [u^k,i]の転置行列TU^と、対角要素の総和が1である非負対角行列R^ = diag [r^k]と、各行の総和が1であるK行M列の非負行列V^ = [v^k,j]と、の積TU^ R^ V^に近似分解をする。出力部105は、基底パターンkに分類される移動の割合として、r^kを、地点iから出発する確率として、u^k,iを、地点jに到着する確率として、v^k,jを、それぞれ出力する。【選択図】図1