dfm-python 0.5.60__tar.gz → 0.6.0__tar.gz

{dfm_python-0.5.60 → dfm_python-0.6.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dfm-python
-Version: 0.5.60
+Version: 0.6.0
 Summary: Dynamic Factor Model (DFM) estimation and forecasting in Python
 Project-URL: Homepage, https://github.com/yourusername/dfm-python
 Project-URL: Documentation, https://github.com/yourusername/dfm-python#readme

dfm_python-0.6.0/config/dfm_exchange.yaml

@@ -0,0 +1,18 @@
+# DFM Configuration for Exchange Rate Datasets
+# Clock: Daily (d) - base frequency for latent factors
+# Data: Daily exchange rate data (all series at same frequency)
+
+# EM Algorithm Parameters
+# NOTE: Exchange tutorial is long daily data (T~7.6k). Full EM can take hours.
+# Keep tutorial runs short by capping iterations and loosening convergence.
+threshold: 1.0e-04  # EM convergence threshold (tutorial-friendly)
+max_iter: 50  # Maximum EM iterations (tutorial-friendly)
+
+# Data Handling
+clock: d  # Daily clock frequency (base frequency for latent factors)
+scaler: standard  # Scaler for all series: 'standard', 'robust', 'minmax', 'maxabs', 'quantile', or null
+# Note: scaler is handled internally by the model, not set on config object
+# Note: nan_method and nan_k are internal constants (not user-configurable)
+num_factors: 2  # Number of factors
+
+

dfm_python-0.6.0/config/dfm_finance.yaml

@@ -0,0 +1,104 @@
+# DFM Configuration for Financial Datasets
+# Based on coded financial indicators (D, E, I, M, P, S, V series)
+# Data: 97 series (D1-D9, E1-E20, I1-I9, M1-M18, P1-P13, S1-S12, V1-V13, plus targets)
+# Target: market_forward_excess_returns
+# Excluded: risk_free_rate, forward_returns
+# Reduced series for testing purposes
+
+# EM Algorithm Parameters
+threshold: 1.0e-05  # EM convergence threshold
+max_iter: 50  # Maximum EM iterations
+
+# Data Handling
+clock: d  # Daily clock frequency (finance.csv uses daily index via date_id)
+scaler: standard  # Scaler for all series: 'standard', 'robust', 'minmax', 'maxabs', 'quantile', or null
+# Note: scaler is handled internally by the model, not set on config object
+# Note: nan_method and nan_k are internal constants (not user-configurable)
+
+# Block Structure for Financial Data (testing configuration)
+# 4 blocks, 2 factors per block (reduced from 3 for faster testing)
+# This still verifies all block mechanisms while reducing state dimension
+blocks:
+  Block_ALL:
+    num_factors: 2  # Reduced from 3 for faster testing
+    series:
+      # Contains all series from Block_EI, Block_MD, and Block_PSV
+      # From Block_EI:
+      - E1
+      - E5
+      - E10
+      - E15
+      - E20
+      - I1
+      - I5
+      - I9
+      # From Block_MD:
+      - M1
+      - M5
+      - M10
+      - M15
+      - M18
+      - D1
+      - D5
+      - D9
+      # From Block_PSV:
+      - P1
+      - P5
+      - P10
+      - P13
+      - S1
+      - S5
+      - S10
+      - V1
+      - V5
+      - V10
+      - market_forward_excess_returns
+
+  Block_EI:
+    num_factors: 2  # Reduced from 3 for faster testing
+    series:
+      # E series: ~6 series
+      - E1
+      - E5
+      - E10
+      - E15
+      - E20
+      # I series: ~4 series
+      - I1
+      - I5
+      - I9
+
+  Block_MD:
+    num_factors: 2  # Reduced from 3 for faster testing
+    series:
+      # M series: ~6 series
+      - M1
+      - M5
+      - M10
+      - M15
+      - M18
+      # D series: ~4 series
+      - D1
+      - D5
+      - D9
+
+  Block_PSV:
+    num_factors: 2  # Reduced from 3 for faster testing
+    series:
+      # P series: ~4 series
+      - P1
+      - P5
+      - P10
+      - P13
+      # S series: ~3 series
+      - S1
+      - S5
+      - S10
+      # V series: ~3 series
+      - V1
+      - V5
+      - V10
+      - market_forward_excess_returns
+
+# Frequency mapping: Not required for single-frequency cases
+# The code will automatically create frequency dict from data columns using clock frequency (d)

{dfm_python-0.5.60 → dfm_python-0.6.0}/config/dfm_macro.yaml

@@ -6,7 +6,7 @@
 
 # EM Algorithm Parameters
 threshold: 1.0e-05  # EM convergence threshold
-max_iter: 5000  # Maximum EM iterations
+max_iter: 50  # Maximum EM iterations
 
 # Numerical Stability Parameters
 clip_ar_coefficients: true
@@ -27,52 +27,53 @@ scaler: standard  # Scaler for all series: 'standard', 'robust', 'minmax', 'maxa
 # Note: scaler is handled internally by the model, not set on config object
 # Note: nan_method and nan_k are internal constants (not user-configurable)
 
+# Tent weights for mixed-frequency aggregation
+# 3-period tent kernel for monthly->weekly aggregation
+tent_weights:
+  m:w: [1, 2, 1]  # 3-period tent kernel for monthly->weekly aggregation
+
 # Block Structure for Macroeconomic Data
-# Simplified for faster testing - reduced to 2 blocks with essential series
-# Series can appear in multiple blocks to capture different factor structures
+# Single global block for all series
+# Series order: clock-frequency (weekly) first, then slower-frequency (monthly)
 blocks:
-  Block_Real:
-    num_factors: 2  # Real economy factors (production, employment, consumption, investment)
-    series:
-      - KOCONPRCF  # Production
-      - KOEMPTOTO  # Total employment
-      - KOHWRWEMP  # Hours worked
-      - KOWRCCNSE  # Consumption
-      - KOIMPCONA  # Investment
-  Block_Prices_Financial:
-    num_factors: 2  # Price, financial, and survey factors (combined for testing)
+  Block_Global:
+    num_factors: 3  # Number of factors for global block
     series:
-      - KOCPCOREF  # Consumer prices
-      - KOCPEXFEF  # Export prices
-      - KOWRCCNSE  # Consumption (overlap with Block_Real)
-      - KOCREU..A  # Credit
-      - KOLRLTHH  # Interest rates
-      - KOBSIHAPR  # Business survey
-      - KOCNFCONR  # Confidence
+      # Weekly series (clock frequency) - must come first
+      - KORFBEV  # Retail, Food & Beverages (weekly)
+      - KORWHLS  # Retail, Wholesale (weekly)
+      - KORMCMP  # Retail, Miscellaneous & Comparison (weekly)
+      - KORELEC  # Retail, Electronics (weekly)
+      - KORCNST  # Retail, Construction (weekly)
+      - KORMACH  # Retail, Machinery (weekly)
+      - KOROCOINF  # Retail, Other Consumer & Information (weekly)
+      # Monthly series (slower than clock) - must come after clock-frequency series
+      - KOEMPTOTO  # Total employment (monthly)
+      - KOHWRWEMP  # Hours worked (monthly)
+      - KOWRCDURE  # Durable goods consumption (monthly)
+      - KOCONPRCF  # Production (monthly)
+      - KOEQUIPTE  # Investment, Equipment (monthly, target series)
+      - KOWRCCNSE  # Consumption (monthly, target series)
+      - KOIMPCONA  # Investment, Construction (monthly, target series)
 
 # Frequency mapping for mixed-frequency macro data
 # Clock is weekly (w), so monthly series are slower frequency
 # Grouped format: frequencies are specified as lists grouped by frequency code
-# Reduced set for faster testing (~50% reduction: 18 monthly series kept from original ~36)
 frequency:
   # Monthly series (slower than weekly clock) - typically end-of-month observations
-  # Block_Real series
+  # These are the series used in the tutorial
   m:
     - KOEMPTOTO  # Total employment (monthly)
     - KOHWRWEMP  # Hours worked (monthly)
-    - KOIMPCONA  # Investment, Construction (monthly)
-    - KOWRCCNSE  # Consumption (monthly)
+    - KOWRCDURE  # Durable goods consumption (monthly)
     - KOCONPRCF  # Production (monthly)
-    # Block_Prices_Financial series
-    - KOCPCOREF  # Consumer prices (monthly)
-    - KOCPEXFEF  # Export prices (monthly)
-    - KOCREU..A  # Credit (monthly)
-    - KOLRLTHH  # Interest rates (monthly)
-    - KOBSIHAPR  # Business survey (monthly)
-    - KOCNFCONR  # Confidence (monthly)
-    # Target series
     - KOEQUIPTE  # Investment, Equipment (monthly, target series)
-  # Weekly series (same as clock frequency)
+    - KOWRCCNSE  # Consumption (monthly, target series)
+    - KOIMPCONA  # Investment, Construction (monthly, target series)
+  # Weekly series (same as clock frequency - these appear every week like the clock)
+  # This is a mixed-frequency example: clock is weekly, but monthly series are slower
+  # The tutorial now respects YAML frequencies for proper mixed-frequency handling
+  # These weekly series are available in the data at clock frequency
   w:
     - KORFBEV  # Retail, Food & Beverages (weekly)
     - KORWHLS  # Retail, Wholesale (weekly)