PREDICTOR, METHOD AND PROGRAM

PROBLEM TO BE SOLVED: To accurately predict a behavior of an operator without recalculation of a learning process even when a feature quantity of the operator has a deficit or a kind of the feature quantity has changed.SOLUTION: A recommendation part 30: takes a feature quantity of an operator of a prediction object as an input; selects a prescribed number of operators for whom at least one of a plurality of features indicated by the feature quantity of the operator of the prediction object has the same value from each operator; estimates a probability that the operator of the prediction object selects each potential topic on the basis of a probability that each of the prescribed number of the selected operators selects each potential topic; and predicts a visit place where the operator of the prediction object will visit next on the basis of a probability that the operator of the prediction object selects each potential topic, a current position of the operator of the prediction object, a probability that each visit place is selected next to a certain visit place, and a selection probability of each visit place in each potential topic.SELECTED DRAWING: Figure 3 【課題】動作主の特徴量に欠損がある場合や、特徴量の種類が変化した場合であっても、学習プロセスを再計算することなく、動作主の行動を精度よく予測することができるようにする。【解決手段】推薦部３０が、予測対象の動作主の特徴量を入力とし、各動作主から、予測対象の動作主の特徴量が示す複数の特徴のうちの少なくとも１つが同じ値となる動作主を規定数分だけ選択し、選択した前記規定数分の動作主の各々が各潜在トピックを選択する確率に基づいて、予測対象の動作主が各潜在トピックを選択する確率を推定し、予測対象の動作主が各潜在トピックを選択する確率と、予測対象の動作主の現在位置と、ある訪問場所の次に各訪問場所が選択される確率と、各潜在トピックにおける各訪問場所の選択確率と、に基づいて、予測対象の動作主が次に訪問する訪問場所を予測する。【選択図】図３