To reduce the work burden of an operator and to improve funds operation efficiency without deteriorating service for a user.
A time series data input part 3 inputs time series data on transaction information of automatic transaction devices 11-1n. An environment information input part 4 inputs environment information when transaction is generated and they are stored in a time series information data base 5. A clustering part 6 executes clustering for time series data for prescribed time in the time series information data base 5 and stores the result in a pattern data base 7 with environment information. A pattern retrieval part 9 retrieves an appropriate past time series pattern from the pattern data base 7 based on environment information of a prediction period. A transaction situation prediction part 10 predicts the transaction situation of the prediction period by using the retrieved result and outputs the prediction result in a prediction transaction situation output part 11.
ABE TETSUYA
KONTA KAZUNOBU