Excel Export Features
Overview
MixModeler's Excel export functionality creates comprehensive, professionally formatted Excel files containing all aspects of your marketing mix model. These exports are designed for sharing with stakeholders, archival purposes, and model reimport for future analysis.
Exporting a Model
How to Export
Navigate to Model Library
Locate your model in the list
Click the Export button (download icon)
Excel file downloads automatically to your browser's download folder
File Naming: ModelName_Export_YYYYMMDD_HHMMSS.xlsx
Example: TV_Digital_Model_Export_20250104_143022.xlsx
Export Time
Small models (<30 variables): Instant (<1 second)
Medium models (30-100 variables): 1-3 seconds
Large models (100+ variables): 3-8 seconds
Includes decomposition: Add 2-5 seconds
Excel File Structure
Each exported Excel file contains multiple sheets organized by content type:
Core Sheets (Always Included)
Model Info: Model metadata and settings Data: Full dataset used in the model Coefficients: Complete coefficient information with statistics Model Statistics: Overall model fit metrics Residuals: Actual vs predicted values with residuals Variable Transformations: All variable engineering details
Model-Type Specific Sheets
For OLS Models:
OLS Statistics
OLS Diagnostics
OLS Full Summary
For Bayesian Models:
Bayesian Statistics (with prior and posterior information)
Bayesian Diagnostics (MCMC convergence metrics)
Bayesian Model Metrics (WAIC, LOO, etc.)
Optional Sheets
Decomposition (if requested):
Group Decomposition
Variable Decomposition
Weighted Variables (if present in model):
Weighted Variables detail sheet
Sheet Details
Model Info Sheet
Contains model configuration and metadata:
Model Name
Unique model identifier
TV_Digital_Model
KPI
Dependent variable
Revenue
Model Type
OLS or Bayesian
OLS
Date Range
Time period analyzed
2023-01-01 to 2024-12-31
Observations
Number of data points used
104
Features
Number of independent variables
25
R-squared
Model fit statistic
0.8543
Adjusted R-squared
Adjusted model fit
0.8421
Created Date
When model was built
2025-01-04 14:30:22
Data Sheet
Complete dataset with all variables:
Date column: Time periods
KPI column: Dependent variable values
Feature columns: All independent variables
All observations: Including those filtered by date range
Use Case: Verify data quality, check for anomalies, reimport model
Coefficients Sheet
For OLS Models
const
125.45
22.31
5.62
0.0001
81.52
169.38
TV_Spend
3.25
0.42
7.74
0.0000
2.43
4.07
Digital_Spend
2.18
0.31
7.03
0.0000
1.57
2.79
Interpretation:
Coefficient: Effect size - a 1-unit increase in the variable leads to this change in KPI
Std Error: Uncertainty in coefficient estimate
T-statistic: Coefficient divided by standard error
P-value: Statistical significance (<0.05 typically significant)
CI Lower/Upper: 95% confidence interval bounds
For Bayesian Models
const
Normal
0
100
128.32
21.45
86.84
170.15
1.00
1245
TV_Spend
Normal
0
10
3.18
0.38
2.45
3.91
1.00
1382
Additional Bayesian Fields:
Prior Distribution: Type of prior used (Normal, Student-t, etc.)
Prior Mean/Std: Prior beliefs before seeing data
Posterior Mean/Std: Updated beliefs after seeing data
HDI (Highest Density Interval): 95% credible interval
R-hat: Convergence diagnostic (should be <1.01)
ESS: Effective sample size
Model Statistics Sheet
OLS Models
R-squared
0.8543
Adjusted R-squared
0.8421
F-statistic
45.32
F-statistic p-value
<0.0001
AIC
1245.67
BIC
1289.45
Log-Likelihood
-615.83
Observations
104
Degrees of Freedom (Model)
25
Degrees of Freedom (Residuals)
78
Bayesian Models
Includes MCMC settings and model comparison metrics:
Chains
4
Draws
2,000
Tune
1,000
Target Accept
0.95
Max R-hat
1.002
Min ESS Bulk
1,245
Divergences
0
WAIC
1238.45
LOO
1240.12
Residuals Sheet
Detailed prediction analysis:
2023-01-08
1250.0
1225.5
24.5
24.5
1.96%
2023-01-15
1340.0
1365.2
-25.2
25.2
-1.88%
Uses:
Identify periods with large prediction errors
Check for systematic patterns in residuals
Validate model accuracy
Find outliers or anomalies
Variable Transformations Sheet
Documents all variable engineering:
TV_Adstock_50
adstock
TV_Spend
rate=0.50
AD_50
Digital_Log
log
Digital_Spend
base=e
LOG
TV_x_Season
multiply
TV_Spend × Seasonality
-
MULT
Campaign_Q1
split_by_date
Campaign_Total
start=2024-01-01, end=2024-03-31
SPL_Q1
Benefits:
Complete audit trail of transformations
Reproducibility documentation
Easy reference for similar models
Stakeholder transparency
Working with Exported Files
Opening and Reviewing
Recommended Software:
Microsoft Excel 2016 or later (best compatibility)
Google Sheets (limited formatting)
LibreOffice Calc (good compatibility)
Tips:
Files use standard .xlsx format
All formulas are values only (no Excel formulas)
Safe to share - no macros or executable content
Can be opened on any platform
Editing Exported Files
Safe to Edit:
Add notes or comments
Highlight important values
Create additional sheets for analysis
Add charts or visualizations
Do Not Edit (if planning to reimport):
Sheet names
Column headers
Data structure
Variable names in transformation sheet
Sharing with Stakeholders
Include These Sheets:
Model Info (overview)
Coefficients (key results)
Model Statistics (validation)
Decomposition sheets (insights)
Optional to Include:
Residuals (technical validation)
Full Data (may contain sensitive information)
Variable Transformations (technical details)
Presentation Tips:
Add an Executive Summary sheet at the beginning
Use conditional formatting to highlight key coefficients
Include interpretation notes for non-technical audiences
Consider creating charts from coefficient data
Export Options
Standard Export
Default export includes all core sheets plus model-type specific sheets.
Command: Click "Export" button in Model Library
Contents: Model Info, Data, Coefficients, Statistics, Residuals, Transformations, Type-specific sheets
Export with Decomposition
Includes additional decomposition analysis sheets.
Command: Check "Include Decomposition" before exporting
Additional Sheets:
Group Decomposition (channel category contributions)
Variable Decomposition (individual variable contributions)
Use When: Sharing attribution insights with marketing teams
Selective Sheet Export
Some users may want only specific sheets for different audiences.
Workaround: Export full file, then delete unnecessary sheets before sharing
Common Configurations:
Executive Summary: Model Info + Coefficients + Decomposition
Technical Review: All sheets
Marketing Team: Model Info + Coefficients + Decomposition + Residuals
Finance Team: Model Info + Decomposition (for budget allocation)
File Size and Limits
Typical File Sizes
Small model (20 variables, 52 observations): 150-300 KB
Medium model (50 variables, 104 observations): 400-800 KB
Large model (200 variables, 260 observations): 1.5-3 MB
Very large model (500 variables, 500 observations): 5-10 MB
Browser Download Limits
No practical limits - MixModeler exports work within standard browser download capabilities.
Maximum Tested: 500 variables × 500 observations successfully exported (10 MB file)
Common Use Cases
Archive and Documentation: Export models for record-keeping and compliance
Stakeholder Reporting: Share results with marketing, finance, or executive teams
Model Comparison: Export multiple models to compare side-by-side in Excel
Further Analysis: Use exported data for custom analysis in Excel or other tools
Reproducibility: Document exact model specifications for future reference or audits
Reimport: Save models to reload later (see Model Reimport documentation)
Next Steps: Learn about Model Export Structure to understand file organization, or explore Model Reimport to reload exported models.
Last updated