Load your data
Choose what the baseline solves on: the sample data, your own datasets bound table by table, or a workbook mapped straight into the model.
Where: New Model→Load data
Load data is step 4 of the New Model wizard. It sets the numbers the baseline solves on: the build's sample data, the rows RIA read from data you attached, or every row of your own data.
See what step 4 offers
The intro, Now the numbers, and the first screen change with what the build ran on:
| First screen | When | Choices |
|---|---|---|
| Sample data, or your own? | The build proved the structure on generated sample data | Create the baseline with sample data or Load my own data |
| The data the build used, or more of your own? | The build filled the model from the data you attached | Create the baseline with your dataset, or Load or replace datasets |
| The rows RIA read, or every row? | The build read part of your data | Create the baseline with the rows RIA read ("A sample of" your dataset), or Load every row of your dataset |
| Your model runs on sample data for now | RIA read your data but its numbers were refused, so the build fell back to sample data | What stopped it: lists up to five reasons. Keep the sample, or fix the workbook and Load my own data |
| This model did not ask you for data / Nothing more to load | The build ran on the numbers your description gave | Nothing to choose; Next creates the baseline |
| Map your workbook into the model | You attached a workbook and confirmed its design | See Review a workbook mapped into the model |
While the tables load: Reading the requisition. If that fails, Couldn't read the requisition: reload.
Keep the sample data
Pick Create the baseline with sample data and click Next: the generated numbers stay, and the wizard moves on to Solve. Load your own later from the model's page. Next with no card picked says "Choose sample data or your own."
See the tables the model asks for
Under the cards, The tables lists every table the model reads: "Tables marked your data came from your files; placeholder tables hold sample numbers until you add yours."
| On a table | Means |
|---|---|
| REQUIRED / OPTIONAL | Whether the model needs it. Hover REQUIRED to see which part of the design uses it |
| your data, placeholder, to provide | Came from your files; holds sample numbers; still needs yours |
one row per site | What one row stands for |
| 6 columns | Folds open the columns, with what a blank means |
Excel template and CSV pack download every table as a workbook or CSV files, with example rows: "replace them with your data; quantities load verbatim." Tables the design fills itself fold under 3 tables filled in for you from your design; optional ones under 2 optional tables.
Load every table from your workbook
When you attached a workbook at Describe and the build didn't carry all of it, a Recommended button appears: Load every table from your workbook. "RIA reads every row of every sheet, matches each sheet to the table it feeds and binds them to your model. You then check the binding before the baseline solves." It shows "Reading every row of" your workbook, then "Binding" it, then opens the binding review (Bind and accept), and keeps running if you leave. If it fails, it says why and that nothing was loaded: try again, or load the tables one at a time.
Point each table at a dataset
- Pick Load my own data (or Load table by table) and click Next. Load your own data lists one
row per table, like
Required · 6 columns. - Click Add on a table and pick a dataset under Datasource ("Pick a dataset…"). "Not here yet? Import it in the Dataset Library, then pick it above." opens Datasets in a new tab.
- RIA proposes how it fits (Proposing how this dataset binds). Check it; see below.
- Next returns to the list; repeat for each table you have numbers for. One dataset can feed several.
Talk to RIA about this data, on each table's screen, takes a note on reading the dataset ("The Region column is really the site name"); Send passes it to the RIA console and keeps it with the table. "No data to load yet?" on the list offers Create the baseline with sample data, straight to Solve. Your notes and picks are kept in this browser until you bind.
Check the proposed binding
| Part | What it shows |
|---|---|
| Overall match | A percentage: green from 85%, amber from 50%, red below |
| Counts | "4 of 6 expected tables fed · 1 needs review · 2 on sample data" |
| Model expects / From dataset table | Each expected table and the table of yours that feeds it. Pick another, or — none (keep sample data) — |
| Confidence / Status | How sure the match is. Below 85% a row needs review; otherwise accepted. sample data when nothing feeds it; pending re-stage after you change it, until you re-score. A count such as 2 issues flags a row with findings |
| Accept | Tick to accept a row. Accept all suggestions accepts every row that needs review |
Click a table's name to open why it matched, any issues found with it, and its columns: Expected column, From column (or — unmapped —), Confidence and Type. A column filled by a rule reads "every row:" and the value, or "made from the" columns it's built from; type a value in the box to set it on every row. your pick marks a column you chose; type differs, a type mismatch. A badge says what each column does: key, "used for" and what it feeds, or not loaded. "Not used from" lists columns nothing took; Not used from the dataset: lists sheets nothing took, each with why, such as "Only 2 of its 7 columns match Demand." Place on table points that table at the sheet. After any change, Re-score with these changes scores the binding again.
Bind and accept
- Click Next on the table list. Review your datasources lists each table:
Orders · 92% match · 1 note to RIA, Required · no datasource yet or Optional · skipped. - Click Bind the data to the model (Binding your data to the model).
- Your data, bound to the model: "Check the binding, then solve the baseline on your numbers." Read the validation report, which checks every table you bound, then click Accept and solve.
| Headline | Means | Do |
|---|---|---|
| validation clean | Nothing found | Accept and solve |
| "2 warnings — advisory" | Worth a look; doesn't stop anything | Accept and solve |
| "3 blocking findings" | Rows the build can't use as they are | Fix the data and bind again, or Accept with exceptions: tick that you reviewed them and give a reason ("Why is this data acceptable as-is? (required)"). The reason is recorded |
| "1 fatal finding — nothing buildable" | The build can't run on this data | "The binding has a fatal problem; fix the datasource and bind again." |
Each table shows "7,600 of 7,600 rows from items" and its findings by row and column, with a suggested fix where there is one. Up to 200 findings are listed; the counts are always complete. If the check found something in your data the build has to know first, accepting says "Before the build, answer what the check found in your data": accept the repair each offers, leave that demand out, or revise the workbook.
Load a workbook straight into the model
Your data in one workbook offers Load a workbook (.xlsx): "RIA reads every sheet and shows you how
it read them. Once you agree, your data goes straight into the model, and you check it before the baseline
solves." The reading opens at step 4 with three pages, as at Clarify but without the decision sentence
(Start from a workbook). Its last button,
Map my data into the model, leads to the mapping review below.
| Refused with | Means |
|---|---|
| "file must be .xlsx" | Save it as an Excel workbook |
| "could not be read into a dataset, so the model's data was not changed" | Check the workbook has its data on named sheets |
| "this model already loads its data from a workbook" | Use Load a revised workbook instead |
| "this model's data was loaded as tables, and it keeps that way" | Bind tables as above |
Review a workbook mapped into the model
A model built from a workbook skips the binding. RIA maps your rows straight into the model (Mapping your workbook into the model; this wait can't be left), and step 4 opens on Your data, bound to the model: "Nothing is built yet. Check it, then build and solve on your own rows."
Choose the first slice
When your workbook names two or more business units, Choose your first slice: "The first build solves one business unit of your own rows. You load the rest after it solves." Each unit shows its counts of items, sites and demand rows; the one with the most demand is marked Most demand. Pick another, or All business units, and click Map unit instead. If nothing is mapped yet, Map your workbook into the model offers the same choice with Map unit into the model.
Read the review
| Part | Shows |
|---|---|
| Every sheet of your workbook | Each sheet with Placed by rule, Placed by RIA: check it or Not in the model, and what became of its rows: "238 of your 240 rows are in the model", kept as context, left out by your slice, set aside by you, or rejected. Cells, Columns and Why fold open where the numbers come from, what each column means, and each rejected row with its reason |
| Size line | "Your Orders sheet gives 2,206 rows". A table over 200,000 rows can't be built at once: "Build one business unit at a time: pick a unit above." |
| Sheets not in the model | "The model is built without this sheet, and its result will be too." (after the build, "…and its result is too."). Left out, and why says why for each |
| What your workbook does not say | Values the model needs that your sheets don't give, each tagged Suggested or Your value. Change one and click Map again with your values |
| Gaps in your data | Each finding with its answers as buttons, such as what a unit left short costs (a figure above 0), or which sites are candidates. Answering maps again. Once answered: "Answered by you" and when |
| Notes on your inputs | Under the goal line, a list of how the mapping read what your sheets leave open, such as "Your Customer Demand sheet names its periods its own way; they are read onto the model's in time order: Week 1 as W1." |
| Goal line | "The model is built to meet demand at the least cost. Every stated demand must be met; a shortfall is not allowed." or "The model is built for the most profit. Demand is a ceiling; the model sells what pays." |
| Every row of your workbook | Off the sheet-by-sheet view, each table's rows instead: "Loaded 2,200 of 2,206 rows from Orders; 6 rejected", with Why for the rejected ones |
Build and solve
Click Build and solve on unit (or Build and solve on your data): "Builds with the values shown above." Rejected rows are named above and not loaded. The wizard moves on to Solve. If a baseline was solved before, "Your baseline solved before still stands, and Next opens it at Solve. The data mapped here is not in it yet."
When the mapping stops
The header pill reads STOPPED. When the mapping itself couldn't run, the title is The mapping stopped, with why. When it ran but nothing can be built from it, the title stays Your data, bound to the model: "RIA read" your workbook "against the model, and nothing can be built from it yet: each sheet below says what stops it." Under the sheets: "Correct the workbook, then map it again." Most stops name the fix; Map it again reruns the mapping once you've made it.
| It says | Do |
|---|---|
| What to call things: "What should the model call…" | Pick or type each name, then Map again with these names. Name them on the reading goes back to the reading instead |
| Sheets nothing loaded | Set it aside and map again, or Place it on the reading (Set them aside… / Place them… for several) |
| Columns to pick: "Which column of your Orders sheet holds the…" | Pick the column, then Map again with this column (or these columns) |
| "Your workbook holds only the template's example rows…" | Replace them with your rows, or include the examples as your data |
| "A profit model needs a selling price for everything it sells…" | Put the price on your demand or items sheet, or go back to Describe and choose least cost |
| "A row of your workbook gives a minimum above its maximum…" | Correct those rows |
| "…too large to build as one model…" | Pick one business unit |
| "The dataset this model loads its workbook from is no longer in Datasets." | Load a revised workbook |
| "Nothing in your workbook could be built into the model yet." | Correct the workbook, or its reading, as each sheet says |
When the model's tables can't take your workbook at all, Build it from your workbook as written offers Build it from the workbook: "RIA can build the model from your workbook the way it builds one from a description, reading its sheets as written." The wizard goes back to Build; an earlier baseline stays until the new one solves.
Load a revised workbook
Load a revised workbook (.xlsx) takes a corrected copy: it becomes a new dataset, the model points at
it, and it's mapped again on the same slice. One that adds a sheet opens the reading first; see
Start from a workbook.
Load the rest
After the first slice solves, step 4 says Your data is in the model: "The model is built on the rows of" your workbook "below, and nothing else." If the review after the build found anything, See them in Develop opens it (Change a model). "Built on your data:" names the unit, with its items and sites. Load the rest maps every business unit "through the same mapping. You check it, then the baseline solves again on all of it." The same line and button sit above the Solve screens.
Review imported data on an older workbook model
Some workbook models open step 4 on Review your imported data instead: each imported value checked against the cell it came from, category by category.
- At the top, Your workbook plan vs the optimized model sets your figures (Imported from workbook) beside the model's (Optimized within your limits). Why do these differ? asks RIA to explain, marked Written by RIA or From known patterns; Work it out again reruns it. A line marked from workbook was rebuilt from your own numbers; AI estimate means RIA interpreted part of it, so a large difference there needs your review. With nothing to pair, Headline KPIs says so; the checks below still apply.
- Constraints from your source data lists the caps that bind the plan.
- Each category shows "matched · derived · traced" counts, its Source tables: (sheet and range) and its fields (Imported field, Value, Matched to source); Show n unmatched narrows to the fields that didn't match. Mark it reviewed, or Accept all. "2 of 6 categories accepted" tracks you; a reload clears the ticks.
- Data-validation findings lists warnings by cell; filter all, warn or block. Tick the acknowledgement under them before you can accept.
- Accept Base & promote solves the baseline. Or Send back with a reason (the model returns to its
questions with your answers kept), or Replace the workbook with a corrected
.xlsx.
A model with no mapping to review shows its Model Design and How this is built (for modelers) here instead, with the same Accept Base & promote. The same screen opens from Open reconciliation on Home. If it fails to load, "Couldn't load the imported data." offers Retry.
Load data into a model that's already built
Where: Overview→Suggested next→Load your data
A model still on sample data suggests Load your real data: "The structure is proven on generated test data — your numbers haven't moved in yet." Load your data opens the model's Credentials record, where Load your data… opens two tabs:
- Upload filled template: the Excel template from The tables, filled in.
- Use a dataset: pick one under Dataset ("Choose a dataset…"), then Stage & validate to check it against the model and show the binding and its Data validation. Change the binding and click Re-stage with changes, then Map data onto the verified structure. Accepting the mapping is on Credentials.
A model built from its workbook offers Load a revised workbook… there instead, which opens step 4.
Change the data later
On a solved model, picking another card at step 4 asks Change the data?: the baseline is redone. Change the data confirms.
Limits
- A row binds automatically from 85% confidence.
- Up to 200 findings are listed per report.
- A table can hold up to 200,000 rows for the build.
- Workbooks upload through a 100 MB limit,
.xlsxonly. - One dataset per model.
Related
- Datasets and Import data: bring the data in first
- Start from a workbook: the reading of your sheets
- Build and baseline: the solve
- Edit data: change numbers after the baseline
Review the design
Read the Model Design eode drafts from your description or workbook, confirm it section by section, answer what it asks, and build it or send it back.
Build and baseline
Watch the build, read what it found, recover when it stops, then solve your baseline and check it reads like your business.