<img class=”alignnone size-medium wp-image-909″ title=”Forecast” src=”http://spplan.org/wp-content/uploads/2009/06/forecast1.jpg?w=300″ alt=”Forecast” width=”300″ height=”225″ /><strong>Different Approaches</strong><span style=”color:#000000;”>There are several different approaches to generating the forecast in MCA SPO. Each Location-Part combination can be configured to follow one of these approaches:</span><ol style=”list-style-type:decimal;”><li><span style=”color:#000000;”>Computed: Based upon historical demand and leading indicators such as causals and installed population. Blends the history based and causal forecasts. The blending of these two forecast streams is called composite forecasting</span></li><li><span style=”color:#000000;”>Replacement: Forecasts are computed as a function of parent Part forecasts using replacement rates</span></li><li><span style=”color:#000000;”>External: Forecasts are uploaded from an external system</span></li></ol><span style=”color:#000000;”>These are defined in the tables (Computed, Replacement, External) in the DEMAND_FORECAST_TYPE field in the LP_ATTRIBUTE interface.</span><span style=”color:#000000;”><strong>LP ATTRIBUTE Table</strong></span><span style=”color:#000000;”>LP_ATTRIBUTE fields include:</span><ol style=”list-style-type:decimal;”><li><span style=”color:#000000;”>Location</span></li><li><span style=”color:#000000;”>Part</span></li><li><span style=”color:#000000;”>Condemnation Rate</span></li><li><span style=”color:#000000;”>Passup Rate</span></li><li><span style=”color:#000000;”>Criticality</span></li><li><span style=”color:#000000;”>Variance to Mean Ratio</span></li></ol><span style=”color:#000000;”>The types of forecasts include</span><ol style=”list-style-type:decimal;”><li><span style=”color:#000000;”>Moving average</span></li><li><span style=”color:#000000;”>Constant with smoothing parameters (parameters can be set in the PARAMETER table) however overrides can be provided using the Forecast Detail View</span></li><li><span style=”color:#000000;”>MCA smoothing forecast is a variant of the single exponential smoothing forecast tailored specifically for sparse and intermittent demand forecasting.</span></li><li><span style=”color:#000000;”>Trend (factors here are also set in the PARAMETER table)</span></li><li><span style=”color:#000000;”>Seasonality Forecast (factors here are also set in the PARAMETER table</span></li><li><span style=”color:#000000;”>Trend Seasonality (factors here are also set in the PARAMETER table)</span></li></ol><span style=”color:#000000;”>SPO uses a best fit to select the best of the available forecasting methodologies</span><span style=”color:#000000;”>The composite forecasts (Time Series + MCBF/MTBF) using parameters in the PARAMETER table</span><span style=”color:#000000;”>____________________________</span><h2><span style=”color:#3366ff;”>MTBF Development</span></h2><span style=”color:#000000;”>Developing an MTBF is much more work than commonly thought. Because not a lot of companies do causal forecasting, they often do not maintain good data for even one MTBF. However there is another reason for this related to the companies interest in investing in the long term. This is described in this post. </span><a href=”http://spplan.wordpress.com/2007/07/22/service-parts-and-mtbf-forecasting/”>http://spplan.org/2007/07/22/service-parts-and-mtbf-forecasting/</a><span style=”color:#000000;”>There are companies that can help develop and capture MTBF data. See this post to see a description of one. </span><a href=”http://spplan.wordpress.com/2008/12/09/determining-mtbf/”>http://spplan.org/2008/12/09/determining-mtbf/</a><span style=”color:#000000;”>____________________________
</span><span style=”color:#000000;”>Forecasting will be done for reparable parts. The flag for this is set in the PART.IS_REPARABLE field (set to T for reparable) in the PART table. The PART table has a large number of fields that are critical:</span><span style=”color:#000000;”><strong>PART Table
</strong></span><ol style=”list-style-type:decimal;”><li><span style=”color:#000000;”>Part Type</span></li><li><span style=”color:#000000;”>Cost</span></li><li><span style=”color:#000000;”>Begin Date</span></li><li><span style=”color:#000000;”>End Date</span></li><li><span style=”color:#000000;”>Is Planned (T or F)</span></li><li><span style=”color:#000000;”>Is Seasonal (T or F)</span></li><li><span style=”color:#000000;”>Is Reparable (T or F)</span></li><li><span style=”color:#000000;”>Repair Cost</span></li><li><span style=”color:#000000;”>Is Order Policy ROQ Based</span></li><li><span style=”color:#000000;”>Holding Cost Rate</span></li><li><span style=”color:#000000;”>Fixed Order Cost</span></li><li><span style=”color:#000000;”>Decay Rate</span></li><li><span style=”color:#000000;”>Generate New Buy (T or F)</span></li><li><span style=”color:#000000;”>Generate Repair (T or F)</span></li><li><span style=”color:#000000;”>Generate Allocation (T or F)</span></li><li><span style=”color:#000000;”>Generate Transshipment (T or F)</span></li><li><span style=”color:#000000;”>Max Qty Allowable</span></li><li><span style=”color:#000000;”>Max Total TSL</span></li><li><span style=”color:#000000;”>Is Exempt (T or F)</span></li><li><span style=”color:#000000;”>Ignore Weight and Volume (T or F)</span></li><li><span style=”color:#000000;”>Material Class</span></li><li><span style=”color:#000000;”>Is Cannabalizable (T or F)</span></li><li><span style=”color:#000000;”>Partial One Way Supersession</span></li></ol><span style=”color:#000000;”>By default SPO includes demands for non stockable parts such as exempt parts (see Is Exempt field above) or parts with fixed overrides, in Fill Rate Optimization. This causes SPO to increase Target Stock Levels of other stockable parts while trying to meet a Location Fill Rate.</span>