提高人工智能模型的准确性和性能.pptx

1、,Empowering AI Through Time Series Analysis,Salochina Oad,PhD Usxpress.IncS,Agenda,Why do we need time series analysis?What are the key expectations before starting forecasting?How to deal with variation of time series data?How can time series empower AI?,The value of improving the performance of AI

2、 Models,Challenges in detecting and preventing fraudulent activities,Efficiency&performance,Time series,Trend/Seasonal Residuals,Multiplicative TS,Additive TS,Types of time series data,Time series,Pooled/Panel,Time series use cases,Demand forecast How many order a trucking company got this week-Cost

3、 planning,Sales forecastHow much revenue was generated serving specific customer.-Financial outcomes,Roadmap-Forecasting project,Determine Target,Horizon of the forecast,Gather data,Develop a model,Deploy to production,Monitor,Time series pipeline,EDA-Components of a time series,Airline passenger,ED

4、A-Components of a time series,Trend and Seasonality,Differencing Log,Preprocessing-anomalies,SLT plotCheck standard deviationBack fill using mean,Preprocessing-Special events,Create dummy feature,July 4,July 4,July 4,Preprocessing-Feature engineering,Choosing the best technique for modeling,Non-Para

5、metric methods,Parametric methods,Forecasting-Parametric Methods,Assumption:stationarity AdvantagesParametric methods are simpler,easier to understand and interpret resultsRequires smaller amount of dataComputationally inexpensiveLimitationsInability to capture subtle patterns in time series data,St

6、eps to identify a model,Stocks,ApplyNave forecastingPrevious seasonal point,Gather data,Stationary,ACF,Autocorrelation,No,Yes,No,No,Apply transformation,Random walk,Yes,Autocorrelation coef.abruptly non sig after lag q,MA,Yes,No,Moving average(q),Random walk,Not MA,Auto regressive(p),Moving average(