| Oracle® Retail Demand Forecasting User Guide for the RPAS Fusion Client Release 16.0.1 E89277-05 |
|
 Previous |
 Next |
| Oracle® Retail Demand Forecasting User Guide for the RPAS Fusion Client Release 16.0.1 E89277-05 |
|
Previous |
Next |
This chapter describes these two features:
Item attributes — which are used to make automatic like item recommendations for new items.
New items — including:
workflow around handling
reviewing and approving their automatic like item recommendation
the manual process for like stores selection with new stores
In general, an item is eligible to be considered new if it satisfies the following condition:
Forecast start date override is in the future or Sales history length is less than the time series duration parameter
In general, an item is eligible to be considered a like-item if its recent sales density is acceptable.
These workbooks help you achieve this task:
The following table lists the workbooks, steps, and views for the New Item & Locations task.
|
Note: The full functionality of the New Item Maintenance workbook is available only when item attributes are loaded and thus the like item recommendation is automated.When item attributes are not available the like item has to be selected manually. However, the workbook is the same for both attribute and non-attribute cases. If at a later point in time, attributes become available, the automated like item recommendation can be used, without the need to patch the environment. |
To build the New Item Maintenance workbook, perform these steps:
Click the New Workbook icon in the New Item Maintenance task in the New Item & Locations activity.
The Workbook wizard opens. Select the domain for this workbook and click OK.
The Workbook wizard opens. Select the products you want to work with and click Next.
Select the locations you want to work with and click Finish.
The New Item Maintenance workbook is built.
This step contains views that allow you to review and approve like item recommendations, when item attributes are available. If attributes are not available, the like item selection is done manually. You also have visibility to which items are eligible to serve as like items for the new items by location.
The available views are:
The Select and Approve view is used to reviews the system generated like item recommendations if attributes are available. You can overwrite the recommendations, and/or trigger the approval, at the granular item / store intersection. This view has a custom menu, Approve New Items, and it approves the like items recommendations for new items. The recommendations are the system recommended or overrides, depending on the approve settings.
The Select and Approve view contains the following measures:
Approve Date
This measure displays the date when the like item recommendation was approved by running the custom menu.
Approve
This measure determines which recommendations are approved the next time the custom menu is run. The options are:
Null— no like item is recommended.
Approve System — the system recommended like item is approved.
Approve Override — your selected like item is approved. Note that if no user selections are available, then no like item is approved.
Substitute Method
This measure displays a list where you can select the substitute method. When a Substitute Method is used to forecast, the method set for an intersection is cleared once the Default Forecast Start Date is greater than the Forecast Start Date Override plus the Like TS Duration for the intersection. Valid options are:
| Substitute Method | Description |
|---|---|
| None | No Forecast is created for the time series (product location combination) |
| Seasonal | You provide a like item/location that has a similar seasonality pattern. The new time series' forecast is the like item/locations demand forecast with the applied adjustment. The forecast is set to zero (0) for all dates before the new product/location's start date. |
| Lifecycle | You provide a like item/location that had a similar lifecycle pattern as the new item/location. The new item/location's forecast is the like/location's actual historical demand with the applied adjustment shifted such that the like item/locations first sales matches the new item/location's forecast start date. |
| Cloning | You provide a like item/location that has a similar selling pattern as the new item/location. The historical demand of the like item/location is copied into the historical sales of the new item/location. During forecasting, the forecast of the new item/location is generated based on the new item/location's own, copied historical demand. |
| Base Rate of Sales | You provide a base rate of sales for a new item/store combination. The new product's forecast is a combination of the seasonality of the forecast at the corresponding source level and the base rate of sales. Specifically, the formula to calculate the forecast is:
Forecast at time t = source level forecast at time t divided by the average of the source level forecast times the base rate of sales. The base rate of sale is a convenient way to generate (seasonal) forecasts for a new item. No Like item or Clone item is needed. What is necessary is a value for how much the item is selling on average per period. That value works as the interim forecast used to spread the source level forecast down to final level. Because this item is new, the system knows that as you have specified a history start date. The base rate of sale forecast is generated until the TS duration is reached. The TS duration is the difference between forecast start date and history start date. Usually the default history start date is empty, which means it is the beginning of the calendar. If the history start date override is also empty, the forecast start date less the beginning of the calendar (>=2 years) is probably much more than the TS duration, so the item is not considered new. In this case base rate of sales or any new item method, will not work. |
Percent Contribution of Like Item 1
This measure determines the percentage of Like item 1's forecast that is going to be applied towards the forecast of the new item.
Percent Contribution of Like Item 2
This measure determines the percentage of Like item 2's forecast that is going to be applied towards the forecast of the new item.
Percent Contribution of Like Item 3
This measure determines the percentage of Like item 3's forecast that is going to be applied towards the forecast of the new item.
Adjustment Factor
This measure determines how much the combined forecasts are scaled up or down to create the forecast of the new item.
System Recommended Like Item 1
This measure displays the top like item for a given store.
System Recommended Like Item 2
This measure displays the second like item for a given store.
System Recommended Like Item 3
This measure displays the third like item for a given store.
User Selected Like-Item 1
This measure allows you to override the top like item for a given store.
User Selected Like-Item 2
This measure allows you to override the second like item for a given store.
User Selected Like-Item 3
This measure allows you to override the third like item for a given store.
Base Rate of Sales
This measure represents the average sales of a new item/store combination. It is used when specifying Base Rate of Sales New SKU as the Substitution Method to create a forecast for the new item/store combination. The measure can be generated in another application and loaded into RDF, or it can be manually entered by a user.
Forecast Start Date Override
This measure represents the date to start forecasting for an item/location combination. This measure can be set in the future if using like-item or Sister-Store functionality, and, upon reaching that time, the forecast is generated. If this date is set to the past, it is ignored in favor of the Forecast Start Date from the Forecast Administration Workbook. This means that the Forecast Start Date for this intersection needs to be edited once it is no longer in the future. For like-item or sister store, the Forecast Start Date and the History Start Date should be set to the same date. It is important to understand how Forecast Start Date should be used in conjunction with Forecast End Date. No value is in this measure if the system default set in the Forecast Administration Workbook is to be used.
|
Note: This measure can also be set in the Forecast Maintenance Workbook. Changes to this measure can be seen in the Forecast Maintenance task. The most recent commit (between either of the tasks) is the value used by the system. |
The Eligible Like Item view is used to display what items are eligible to be like items for each new item included in the view
This step contains views that allow you to set default values for some parameters related to the new item functionality.
The available views are:
The Clone Adjustment Parameters view is used to review default values for some parameters related to the new item functionality.
The Clone Adjustment Parameters view contains the following measures:
Cloned History Adjustment - Alpha (Range 0-1)
This measure is used to calculate the ratio that is applied to the cloned history. If this parameter is close to zero (0), then the cloned sales will be scaled so that they are aligned with the sales level. If this parameter is close to one (1), then little scaling of the cloned history will occur.
Cloned History Adjustment - Recent Sales Threshold
This measure represents a specific number of periods. If it is less than the number of periods from the first sale date to the current date, then the cloned history is adjusted.
If it is greater than the number of periods from the first sale date to the current date, then no adjustment is made to the cloned history.
Cloned History Adjustment - Calculation Window Length
This measure is the number of periods that are considered when adjusting the cloned history.
The New Item Basic Parameters view is used to set default values for some parameters related to the new item functionality.
The New Item Basic Parameters view contains the following measures:
|
Note: If no item attributes are available, then History Time Series Duration is the only measure for this view. |
Percent Contributions
Percent Contribution1, Percent Contribution2, Percent Contribution3
The percent contribution measures determine the percentage of Like Items' forecast that is going to be applied towards the forecast of the new item.
Adjustment Factor
You may enter an Adjustment Factor to apply to the cloned history of the new location. This is a real number between [zero (0), infinity). The default (NA) value is 1.00 (in other words 100%), which translates to no adjustment.
Example 4-1 Adjustment Factor
If demand for a new store is expected to be 30% greater than its clone store, the Adjustment Factor would be set to 1.30. If demand for a new store is expected to be 30% less than its clone store, the Adjustment percent would set to 0.70.
|
Note: Adjustment Factors and Clone Contributions specified in the Product Cloning and Location Cloning views are used together while evaluating the result for a given item/store. |
Example 4-2 Adjustment Factors and Clone Contributions
| SKU1 | |
|---|---|
| SKU2 | SKU3 |
| 20% | 80% |
| Adjustment Factor = 1 | |
| STR1: | |
|---|---|
| STR2 | STR3 |
| 50% | 50% |
| Adjustment Factor = 0.5The contributions are calculated as:SKU2/STR2 at 5% (=20% x 0.5x50%), SKU2/STR3 at 5%,SKU3/STR2 at 20%(=80%x0.5x50%),SKU3/STR3 at 20% | |
Auto-Approve
You decide if automatic like item recommendations are automatically approved by selecting this measure. Automatically approved like items also trigger an alert and the New Item Maintenance workbook is using the alert to pre-range the new items only to approved items positions. If the automatic like item recommendations are not automatically approved, the new items are alerted in a not approved alert and can be viewed and adjusted in the New Item Review workbook.
Substitute Method
This measure displays a list where you can select the substitute method. When a Substitute Method is used to forecast, the method set for an intersection is cleared once the Default Forecast Start Date is greater than the Forecast Start Date Override plus the Like TS Duration for the intersection. Valid options are:
| Substitute Method | Description |
|---|---|
| None | No Forecast is created for the time series (product location combination) |
| Seasonal | You provide a like item/location that has a similar seasonality pattern. The new time series' forecast is the like item/locations demand forecast with the applied adjustment. The forecast is set to zero (0) for all dates before the new product/location's start date. |
| Lifecycle | You provide a like item/location that had a similar lifecycle pattern as the new item/location. The new item/location's forecast is the like/location's actual historical demand with the applied adjustment shifted such that the like item/locations first sales matches the new item/location's forecast start date. |
| Cloning | You provide a like item/location that has a similar selling pattern as the new item/location. The historical demand of the like item/location is copied into the historical sales of the new item/location. During forecasting, the forecast of the new item/location is generated based on the new item/location's own, copied historical demand. |
| Base Rate of Sales | You provide a base rate of sales for a new item/store combination. The new product's forecast is a combination of the seasonality of the forecast at the corresponding source level and the base rate of sales. Specifically, the formula to calculate the forecast is:
Forecast at time t = source level forecast at time t divided by the average of the source level forecast times the base rate of sales. The base rate of sale is a convenient way to generate (seasonal) forecasts for a new item. No Like item or Clone item is needed. What is necessary is a value for how much the item is selling on average per period. That value works as the interim forecast used to spread the source level forecast down to final level. Because this item is new, the system knows that as you have specified a history start date. The base rate of sale forecast is generated until the TS duration is reached. The TS duration is the difference between forecast start date and history start date. Usually the default history start date is empty, which means it is the beginning of the calendar. If the history start date override is also empty, the forecast start date less the beginning of the calendar (>=2 years) is probably much more than the TS duration, so the item is not considered new. In this case base rate of sales or any new item method, will not work. |
Threshold Recent Sales Density
The value of this measure decides if an item can be recommended as like item for new items. The idea behind determining the eligibility is that an item needs to be actively selling to be eligible. We don't want a stale item, or item with no sales to be selected as like item even if item attributes match very well. To determine the eligibility we first calculate the count of non-zero sales in the most recent periods given by the TS duration. Then we divide the count by the TS duration to get the sales density. Finally, the density is compared with the value of the threshold. If it is larger, then the item can be assigned as like item.
History Time Series Duration
In this measure you can enter the threshold for the historical demand duration to determine if a time series is considered. If the demand history is less than the threshold, the forecast of the new time series is generated using the new item functionality. If the demand length is larger than the threshold, the time series is not considered new anymore, and its own demand is used to generate the forecast.
Number of Periods to Alert for New Identified Items
This measure defines how many periods of sales an item has to be alerted as a new item. For instance, if the value is 5, and the item has 4 weeks of sales, it is still considered new, and the alert is triggered.
0,2,3,1,2,1,0,1,2,2,1,today
The calculated number of consecutive periods with positive demand is 4, and the item/location qualifies as a possible like item/location.
Use this workbook to assign like stores to handle forecasting for new stores. The like store assignment is manual and there is a good reason for it. New stores have a large financial impact, so it makes sense having a business person making the like store selection.
It is probably more appropriate than going with an automatic selection based on something like store attributes. Also, new store introductions are infrequent compared to new item introductions, so manually handling new stores is not a significant overhead activity.
To build the New Store Maintenance workbook, perform these steps:
Click the New Workbook icon in the New Store Maintenance task in the New Item & Locations activity.
The Workbook wizard opens. Select the domain for this workbook and click OK.
The Workbook wizard opens. Select the locations you want to work with and click Finish.
The New Store Maintenance workbook is built.
This step contains the Product Like Store Assignment View.
The Product Like Store Assignment view is at the intersection of prod/location, so all parameters can vary by product. For example, a new store opening in the Midwest can have a Like Store from Alaska for items in the Shovels department. However, for summer items, the Like Store is picked from the Northeast region.
The Product Like Store Assignment view contains the following measures:
Like Store 1
In this measure, you can specify the first like store. Note how the like store can be different by product. In RDF, a different first like store selection can be made for every subclass.
Like Store 2
In this measure, you can specify the second like store. Note how the like store can be different by product. In RDF, a different second like store selection can be made for every subclass.
Like Store 3
In this measure, you can specify the third like store. Note how the like store can be different by product. In RDF, a different third like store selection can be made for every subclass.
Adjustment Factor
This measure determines how much the combined forecasts are scaled up or down to create the forecast of the new item.
Substitute Method
This measure displays a list where you can select the substitute method. When a Substitute Method is used to forecast, the method set for an intersection is cleared once the Default Forecast Start Date is greater than the Forecast Start Date Override plus the Like TS Duration for the intersection. Valid options are:
| Substitute Method | Description |
|---|---|
| None | No Forecast is created for the time series (product location combination) |
| Seasonal | You provide a like item/location that has a similar seasonality pattern. The new time series' forecast is the like item/locations demand forecast with the applied adjustment. The forecast is set to zero (0) for all dates before the new product/location's start date. |
| Lifecycle | You provide a like item/location that had a similar lifecycle pattern as the new item/location. The new item/location's forecast is the like/location's actual historical demand with the applied adjustment shifted such that the like item/locations first sales matches the new item/location's forecast start date. |
| Cloning | You provide a like item/location that has a similar selling pattern as the new item/location. The historical demand of the like item/location is copied into the historical sales of the new item/location. During forecasting, the forecast of the new item/location is generated based on the new item/location's own, copied historical demand. |
| Base Rate of Sales | You provide a base rate of sales for a new item/store combination. The new product's forecast is a combination of the seasonality of the forecast at the corresponding source level and the base rate of sales. Specifically, the formula to calculate the forecast is:
Forecast at time t = source level forecast at time t divided by the average of the source level forecast times the base rate of sales. The base rate of sale is a convenient way to generate (seasonal) forecasts for a new item. No Like item or Clone item is needed. What is necessary is a value for how much the item is selling on average per period. That value works as the interim forecast used to spread the source level forecast down to final level. Because this item is new, the system knows that as you have specified a history start date. The base rate of sale forecast is generated until the TS duration is reached. The TS duration is the difference between forecast start date and history start date. Usually the default history start date is empty, which means it is the beginning of the calendar. If the history start date override is also empty, the forecast start date less the beginning of the calendar (>=2 years) is probably much more than the TS duration, so the item is not considered new. In this case base rate of sales or any new item method, will not work. |
Percent Contribution of Like Store 1
This measure determines the percentage of Like Store 1's forecast that is going to be applied towards the forecast of the new item.
Percent Contribution of Like Store 2
This measure determines the percentage of Like Store 2's forecast that is going to be applied towards the forecast of the new item.
Percent Contribution of Like Store 3
This measure determines the percentage of Like Store 3's forecast that is going to be applied towards the forecast of the new item.
Forecast Start
This measure specifies the first date for which forecast is generated for the new item.
This step contains views that allow you to set default values for some parameters related to the new item functionality.
The available views are:
The Clone Adjustment Parameters view is used to review default values for some parameters related to the new item functionality.
The Clone Adjustment Parameters view contains the following measures:
Cloned History Adjustment - Alpha (Range 0-1)
This measure is used to calculate the ratio that is applied to the cloned history. If this parameter is close to zero (0), then the cloned sales will be scaled so that they are aligned with the sales level. If this parameter is close to one (1), then little scaling of the cloned history will occur.
Cloned History Adjustment - Recent Sales Threshold
This measure represents a specific number of periods. If it is less than the number of periods from the first sale date to the current date, then the cloned history is adjusted.
If it is greater than the number of periods from the first sale date to the current date, then no adjustment is made to the cloned history.
Cloned History Adjustment - Calculation Window Length
This measure is the number of periods that are considered when adjusting the cloned history.
The New Item Basic Parameters view is used to set default values for some parameters related to the new item functionality.
The New Item Basic Parameters view contains the following measures:
|
Note: If no item attributes are available, then History Time Series Duration is the only measure for this view. |
Percent Contributions
Percent Contribution1, Percent Contribution2, Percent Contribution3
Adjustment Factor
You may enter an Adjustment Factor to apply to the cloned history of the new location. This is a real number between [zero (0), infinity). The default (NA) value is 1.00 (in other words 100%), which translates to no adjustment.
Example 4-3 Adjustment Factor
If demand for a new store is expected to be 30% greater than its clone store, the Adjustment Factor would be set to 1.30. If demand for a new store is expected to be 30% less than its clone store, the Adjustment percent would set to 0.70.
|
Note: Adjustment Factors and Clone Contributions specified in the Product Cloning and Location Cloning views are used together while evaluating the result for a given item/store. |
Example 4-4 Adjustment Factors and Clone Contributions
| SKU1 | |
|---|---|
| SKU2 | SKU3 |
| 20% | 80% |
| Adjustment Factor = 1 | |
| STR1: | |
|---|---|
| STR2 | STR3 |
| 50% | 50% |
| Adjustment Factor = 0.5The contributions are calculated as:SKU2/STR2 at 5% (=20% x 0.5x50%), SKU2/STR3 at 5%,SKU3/STR2 at 20%(=80%x0.5x50%),SKU3/STR3 at 20% | |
Auto-Approve
You decide if automatic like item recommendations are automatically approved by selecting this measure. Automatically approved like items also trigger an alert and the New Item Maintenance workbook is using the alert to pre-range the new items only to approved items positions. If the automatic like item recommendations are not automatically approved, the new items are alerted in a not approved alert and can be viewed and adjusted in the New Item Review workbook.
Substitute Method
This measure displays a list where you can select the substitute method. When a Substitute Method is used to forecast, the method set for an intersection is cleared once the Default Forecast Start Date is greater than the Forecast Start Date Override plus the Like TS Duration for the intersection. Valid options are:
| Substitute Method | Description |
|---|---|
| None | No Forecast is created for the time series (product location combination) |
| Seasonal | You provide a like item/location that has a similar seasonality pattern. The new time series' forecast is the like item/locations demand forecast with the applied adjustment. The forecast is set to zero (0) for all dates before the new product/location's start date. |
| Lifecycle | You provide a like item/location that had a similar lifecycle pattern as the new item/location. The new item/location's forecast is the like/location's actual historical demand with the applied adjustment shifted such that the like item/locations first sales matches the new item/location's forecast start date. |
| Cloning | You provide a like item/location that has a similar selling pattern as the new item/location. The historical demand of the like item/location is copied into the historical sales of the new item/location. During forecasting, the forecast of the new item/location is generated based on the new item/location's own, copied historical demand. |
| Base Rate of Sales | You provide a base rate of sales for a new item/store combination. The new product's forecast is a combination of the seasonality of the forecast at the corresponding source level and the base rate of sales. Specifically, the formula to calculate the forecast is:
Forecast at time t = source level forecast at time t divided by the average of the source level forecast times the base rate of sales. The base rate of sale is a convenient way to generate (seasonal) forecasts for a new item. No Like item or Clone item is needed. What is necessary is a value for how much the item is selling on average per period. That value works as the interim forecast used to spread the source level forecast down to final level. Because this item is new, the system knows that as you have specified a history start date. The base rate of sale forecast is generated until the TS duration is reached. The TS duration is the difference between forecast start date and history start date. Usually the default history start date is empty, which means it is the beginning of the calendar. If the history start date override is also empty, the forecast start date less the beginning of the calendar (>=2 years) is probably much more than the TS duration, so the item is not considered new. In this case base rate of sales or any new item method, will not work. |
Threshold Recent Sales Density
The value of this measure decides if an item can be recommended as like item for new items. The idea behind determining the eligibility is that an item needs to be actively selling to be eligible. We don't want a stale item, or item with no sales to be selected as like item even if item attributes match very well. To determine the eligibility we first calculate the count of non-zero sales in the most recent periods given by the TS duration. Then we divide the count by the TS duration to get the sales density. Finally, the density is compared with the value of the threshold. If it is larger, then the item can be assigned as like item.
History Time Series Duration
In this measure you can enter the threshold for the historical demand duration to determine if a time series is considered. If the demand history is less than the threshold, the forecast of the new time series is generated using the new item functionality. If the demand length is larger than the threshold, the time series is not considered new anymore, and its own demand is used to generate the forecast.
Number of Periods to Alert for New Identified Items
This measure defines how many periods of sales an item has to be alerted as a new item. For instance, if the value is 5, and the item has 4 weeks of sales, it is still considered new, and the alert is triggered.
0,2,3,1,2,1,0,1,2,2,1,today
The calculated number of consecutive periods with positive demand is 4, and the item/location qualifies as a possible like item/location.
This workbook is intended to review like item recommendations, as well as metrics that support the recommendations. The recommendations are driven by similarity among items, which in turns is based on how close a new item's attributes are compared to all existing items' attributes.
The workbook can be built by manually selecting the new items during the wizard process, or the new item selection can be based on the available new item alert.
To build the Attribute Maintenance workbook, perform these steps:
Click the New Workbook icon in the Attribute Maintenance task in the New Item & Locations activity.
The Workbook wizard opens. Select the domain for this workbook and click OK.
The Workbook wizard opens. Select the products you want to work with and click Next.
Select the locations you want to work with and click Finish.
The Attribute Maintenance workbook is built.
This step contains views that allow you to review attributes for new and existing items. Also, they show how much alike items are and the best choices for like items..
The available views are:
The Attribute Match view is used to review the attributes of new and existing items and how well they match.
The Attribute Match view contains the following measures:
Attribute Value
This measure displays the content of the attribute. For instance the unit of measure (UOM) attribute can have different values. It can be inch or XL for fashion items. Or it can be ounce or grams for grocery items.
Attribute Scores
This measure displays the quantitative fit of the attribute values between new and existing items. For instance we can compare the UOM attribute between a coffee pack and a pair of jeans. The relevance of matching ounces and inches may not be very high, and the attribute score is likely zero. However, if we compare the color attribute of a shirt and a t-shirt, the match be more relevant and the score is larger than zero.
This measure incorporates the goodness of the fit in attributes between new and existing items, as well as how important an attribute is for the new item. If the new item is a carbonated drink, the brand attribute may be much more relevant than the price tier, because the consumer is prepared to pay a higher price for a brand item.
The New Item Attributes view displays attribute information about new items.
The New Item Attributes view contains the following measures:
Attribute Weight
This measure displays the relative importance of the attributes for a given new item. While the flavor family may not be important for a t-shirt, and the attribute has a weight of zero, the brand and color attributes definitely are. Their relative importance may be 0.2 for the color and 0.3 for the brand.
|
Note: That the sum of all attribute weights does not need to be 1 for every item. The automatic like item recommendation algorithm is taking care of it. The most likely scenario is that this measure is loaded, but it can be review and adjusted in this view. |
Attribute Value
This measure displays the content of the attribute. For instance the unit of measure (UOM) attribute can have different values. It can be inch or XL for fashion items. Or it can be ounce or grams for grocery items.
The Existing Item Threshold view is used to adjust the parameter that decides if an item is eligible to be selected as like item for a new item.
The Existing Item Threshold view contains the following measure:
Threshold Recent Sales Density
The value of this measure decides if an item can be recommended as like item for new items. The idea behind determining the eligibility is that an item needs to be actively selling to be eligible. We don't want a stale item, or item with no sales to be selected as like item even if item attributes match very well. To determine the eligibility we first calculate the count of non-zero sales in the most recent periods given by the TS duration. Then we divide the count by the TS duration to get the sales density. Finally, the density is compared with the value of the threshold. If it is larger, then the item can be assigned as like item.
This step contains views that allow you to review the top choices for like items for the new items, based on similarity among items.
The available views are:
The Aggregate Level view is at the item/item RHS intersection. The item represents the new items, while the item RHS represents the like items. The item RHS dimension has only positions that were identified as like items for new items.
The Aggregate Level view contains the following measure:
Aggregated Store Count
This measure displays the number of stores at which an item RHS position was identified as the most suitable like item for a new item. For instance, for new item A, item BA was identified to be the like item for 20 stores. For the rest of 42 stores, it was item CA. This can happen simply because the best fit - in this case item BA - is not sold in all stores. Item CA is the second best fit, but it's sold in more stores, and thus the subjective better fit.
The Item & Location Recommendation view is at the item / store intersection. For new items it displays the top three matching items for a certain store.
The Item & Location Recommendation view contains the following measures:
System Recommended Like Item 1, 2, 3
This measure displays the top three like items (based on similarity and store ranging) for every store. For a given new item, the like items may be different by store. The reason is that for some stores, the existing item with the highest similarity is not available for sale. Then, the algorithm is picking the existing item with the highest score for that store.
The Similarity Score view is at the item/item RHS/attribute intersection. The item dimension contains new item which are manually selected during the wizard process. The new item alert can also be used to range down the items to a relevant selection. The item RHS dimension is ranged to only relevant existing items that are eligible to be like items. The measure shows the calculated similarity between new and existing items.
The Similarity Score view contains the following measure:
Similarity Scores
The Similarity Score is a measure of how well a new item's demand behavior can be modeled after a certain existing item's. The higher the score, the better the fit, and the better the chance that the existing item is going to be the like item
