dfm-python 0.2.6__tar.gz → 0.2.8__tar.gz

{dfm_python-0.2.6 → dfm_python-0.2.8}/.gitignore

@@ -212,4 +212,6 @@ __marimo__/
 *.sh
 
 Nowcasting/
-agent/
+agent/
+
+mybooks/

{dfm_python-0.2.6 → dfm_python-0.2.8}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dfm-python
-Version: 0.2.6
+Version: 0.2.8
 Summary: Dynamic Factor Model (DFM) estimation and nowcasting in Python
 Project-URL: Homepage, https://github.com/yourusername/dfm-python
 Project-URL: Documentation, https://github.com/yourusername/dfm-python#readme
@@ -709,12 +709,20 @@ Demonstrates:
 
 ## Project Status
 
-**Version**: 0.2.6  
+**Version**: 0.2.7  
 **Status**: Stable and production-ready  
 **Python**: 3.10+  
 **PyPI**: https://pypi.org/project/dfm-python/
 
-### Recent Improvements (v0.2.6)
+### Recent Improvements (v0.2.7)
+
+- ✅ **Fixed C normalization issue**: C normalization now only applies when C norm is in reasonable range (0.1 ~ 10)
+  - Prevents Q from becoming too small when C norm is very small
+  - Prevents factor values from becoming extreme
+- ✅ **Improved stability**: Conditional C normalization prevents pathological factor paths
+- ✅ **Better diagnostics**: Added variance guard binding checks in tutorial
+
+### Previous Improvements (v0.2.6)
 
 - ✅ **Q matrix cap**: Factor innovation variances bounded above by 1.0 to prevent explosion
 - ✅ **R matrix cap**: Observation error variances bounded above by 10.0 for standardized data
@@ -758,7 +766,7 @@ If you use `dfm-python` in your research, please cite:
   author = {DFM Python Contributors},
   year = {2025},
   url = {https://pypi.org/project/dfm-python/},
-  version = {0.2.6}
+  version = {0.2.7}
 }
 ```
 

{dfm_python-0.2.6 → dfm_python-0.2.8}/README.md

@@ -663,12 +663,20 @@ Demonstrates:
 
 ## Project Status
 
-**Version**: 0.2.6  
+**Version**: 0.2.7  
 **Status**: Stable and production-ready  
 **Python**: 3.10+  
 **PyPI**: https://pypi.org/project/dfm-python/
 
-### Recent Improvements (v0.2.6)
+### Recent Improvements (v0.2.7)
+
+- ✅ **Fixed C normalization issue**: C normalization now only applies when C norm is in reasonable range (0.1 ~ 10)
+  - Prevents Q from becoming too small when C norm is very small
+  - Prevents factor values from becoming extreme
+- ✅ **Improved stability**: Conditional C normalization prevents pathological factor paths
+- ✅ **Better diagnostics**: Added variance guard binding checks in tutorial
+
+### Previous Improvements (v0.2.6)
 
 - ✅ **Q matrix cap**: Factor innovation variances bounded above by 1.0 to prevent explosion
 - ✅ **R matrix cap**: Observation error variances bounded above by 10.0 for standardized data
@@ -712,7 +720,7 @@ If you use `dfm-python` in your research, please cite:
   author = {DFM Python Contributors},
   year = {2025},
   url = {https://pypi.org/project/dfm-python/},
-  version = {0.2.6}
+  version = {0.2.7}
 }
 ```
 

dfm_python-0.2.8/config/blocks/default.yaml

@@ -0,0 +1,63 @@
+# ============================================================================
+# DFM Block Structure Configuration
+# ============================================================================
+# Each block corresponds to a set of series defined in config/series/default.yaml
+# and represents one latent factor (or more) capturing that sector's dynamics.
+#
+# Each block has:
+#   - factors: number of latent factors in this block
+#              **IMPORTANT**: This should be determined based on:
+#              * Number of series in the block (more series may need more factors)
+#              * Complexity of the sector (simple sectors: 1, complex: 2-3)
+#              * Model fit diagnostics (check factor loadings and explained variance)
+#              * Typical range: 1-3 factors per block
+#              * Start with 1 and increase if model fit is poor
+#   - ar_lag: autoregressive lag order for the block-level factor (typically 1)
+#   - clock: optional, if block frequency differs from global clock
+#   - notes: optional comments
+#
+# **How to determine factors per block:**
+#   1. Start with factors: 1 for each block (simplest model)
+#   2. Check model diagnostics: factor loadings, explained variance per series
+#   3. If many series in a block have low loadings (<0.3), consider increasing factors
+#   4. Use cross-validation or information criteria (AIC/BIC) to compare models
+#   5. Consider economic theory: does the sector have multiple distinct drivers?
+#
+# **Current settings are examples - adjust based on your data and modeling goals!**
+# ============================================================================
+
+Block_Global:
+  factors: 3  # Adjust based on data complexity and model diagnostics
+  ar_lag: 1
+  clock: m
+  notes: "Global (aggregate) factor driving all blocks"
+
+Block_Consumption:
+  factors: 1  # Adjust based on data complexity and model diagnostics
+  ar_lag: 1
+  clock: m
+  notes: "Consumption-related indicators (retail sales, CSI, etc.)"
+
+Block_Investment:
+  factors: 1  # Adjust based on data complexity and model diagnostics
+  ar_lag: 1
+  clock: m
+  notes: "Investment-related indicators (GFCF, BSI Investment, etc.)"
+
+Block_Export:
+  factors: 1  # Adjust based on data complexity and model diagnostics
+  ar_lag: 1
+  clock: m
+  notes: "Export and external demand (goods & services, terms of trade, etc.)"
+
+Block_Labor:
+  factors: 1 # Adjust based on data complexity and model diagnostics
+  ar_lag: 1
+  clock: m
+  notes: "Employment, unemployment, participation, labor expectations"
+
+Block_Nominal:
+  factors: 1  # Adjust based on data complexity and model diagnostics
+  ar_lag: 1
+  clock: m
+  notes: "Nominal price-level or value indicators (PPI, CPI, trade prices, credit, etc.)"