mcp-servers/nadlan-mcp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add: Return outlier deals in analysis responses

Browse Source 
Feature:
- New parameter `include_outlier_deals` (default=True) in:  - filter_deals_for_analysis()
  - calculate_deal_statistics()
  - get_valuation_comparables() MCP tool
  - get_deal_statistics() MCP tool

Behavior:
- When ON: response includes `outlier_deals` field with removed deals
- LLM can see what was filtered out, answer questions about it
- Maintains full transparency on outlier removal

Terminology fixed:
- "outlier_deals" = deals removed as outliers (clearer than "filtered_deals")
- "filtered deals" = deals that PASSED filtering
- Added to OutlierReport model

All 311 tests pass

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
Nitzan Pomerantz
2025-12-06 23:26:21 +02:00
parent ae0ee56bf6
commit 4de1a4822b
5 changed files with 111 additions and 65 deletions
Download Patch File Download Diff File Expand all files Collapse all files
nadlan_mcp/fastmcp_server.py
+77 -58
View File
@@ -799,6 +799,7 @@ def get_valuation_comparables(
radius_meters: int = 100,
max_comparables: int = 50,
iqr_multiplier: Optional[float] = None,
include_outlier_deals: bool = True,
) -> str:
"""Get comparable properties for valuation analysis.
@@ -822,16 +823,19 @@ def get_valuation_comparables(
radius_meters: Search radius in meters (default: 100, larger than find_recent_deals to get more comparables)
max_comparables: Maximum number of deals to return (default: 50, optimized for MCP token limits)
iqr_multiplier: Override IQR multiplier for outlier detection (default: 1.0). Lower = more aggressive filtering
include_outlier_deals: If True (default), include the outlier deals separately in the response
This allows you to see what was filtered out and answer questions about it
Returns:
JSON string containing:
- Filtered comparable deals with full details
- deals: Comparable deals that passed filtering
- deal_breakdown with outlier filtering metadata:
- total_deals: Count after filtering
- total_deals_before_filtering: Count before filtering
- outliers_removed: Number of deals filtered out
- outliers_removed: Number of deals removed as outliers
- filtering_method: Method used (e.g., "iqr")
- iqr_multiplier: IQR multiplier used (e.g., 1.0)
- outlier_deals: List of deals that were removed as outliers (if include_outlier_deals=True)
"""
log_mcp_call(
"get_valuation_comparables",
@@ -905,7 +909,11 @@ def get_valuation_comparables(
):
deals_before_outlier_filter = len(filtered_deals)
filtered_deals, outlier_report = filter_deals_for_analysis(
filtered_deals, config, metric="price_per_sqm", iqr_multiplier=iqr_multiplier
filtered_deals,
config,
metric="price_per_sqm",
iqr_multiplier=iqr_multiplier,
include_outlier_deals=include_outlier_deals,
)
effective_k = (
iqr_multiplier if iqr_multiplier is not None else config.analysis_iqr_multiplier
@@ -936,35 +944,37 @@ def get_valuation_comparables(
if outlier_report["method_used"] == "iqr":
deal_breakdown["iqr_multiplier"] = outlier_report["parameters"]["iqr_multiplier"]
# Normalize response structure to match other tools
return json.dumps(
{
"search_parameters": {
"address": address,
"years_back": years_back,
"radius_meters": radius_meters,
"max_comparables": max_comparables,
"filters_applied": {
"property_type": property_type,
"rooms": f"{min_rooms}-{max_rooms}" if min_rooms or max_rooms else None,
"price": f"{min_price}-{max_price}" if min_price or max_price else None,
"area": f"{min_area}-{max_area}" if min_area or max_area else None,
"floor": f"{min_floor}-{max_floor}" if min_floor or max_floor else None,
},
# Build response with filtered deals
response_data = {
"search_parameters": {
"address": address,
"years_back": years_back,
"radius_meters": radius_meters,
"max_comparables": max_comparables,
"filters_applied": {
"property_type": property_type,
"rooms": f"{min_rooms}-{max_rooms}" if min_rooms or max_rooms else None,
"price": f"{min_price}-{max_price}" if min_price or max_price else None,
"area": f"{min_area}-{max_area}" if min_area or max_area else None,
"floor": f"{min_floor}-{max_floor}" if min_floor or max_floor else None,
},
"market_statistics": {
"deal_breakdown": deal_breakdown,
"price_statistics": stats.price_statistics,
"area_statistics": stats.area_statistics,
"price_per_sqm_statistics": stats.price_per_sqm_statistics,
"property_type_distribution": stats.property_type_distribution,
"date_range": stats.date_range,
},
"deals": strip_bloat_fields(filtered_deals),
},
ensure_ascii=False,
indent=2,
)
"market_statistics": {
"deal_breakdown": deal_breakdown,
"price_statistics": stats.price_statistics,
"area_statistics": stats.area_statistics,
"price_per_sqm_statistics": stats.price_per_sqm_statistics,
"property_type_distribution": stats.property_type_distribution,
"date_range": stats.date_range,
},
"deals": strip_bloat_fields(filtered_deals),
}
# Add outlier deals if present and requested
if include_outlier_deals and outlier_report and "outlier_deals" in outlier_report:
response_data["outlier_deals"] = strip_bloat_fields(outlier_report["outlier_deals"])
return json.dumps(response_data, ensure_ascii=False, indent=2)
except Exception as e:
logger.error(f"Error in get_valuation_comparables: {e}", exc_info=True)
@@ -979,6 +989,7 @@ def get_deal_statistics(
min_rooms: Optional[float] = None,
max_rooms: Optional[float] = None,
iqr_multiplier: Optional[float] = None,
include_outlier_deals: bool = True,
) -> str:
"""Calculate statistical aggregations on deal data for an address.
@@ -992,9 +1003,13 @@ def get_deal_statistics(
min_rooms: Minimum number of rooms
max_rooms: Maximum number of rooms
iqr_multiplier: Override IQR multiplier for outlier detection (default: 1.0). Lower = more aggressive filtering
include_outlier_deals: If True (default), include the outlier deals in the response
This allows you to see what was filtered out
Returns:
JSON string containing statistical metrics (mean, median, percentiles, etc.)
JSON string containing:
- market_statistics: Statistical metrics (mean, median, percentiles, etc.)
- outlier_deals: Deals that were removed as outliers (if include_outlier_deals=True)
"""
log_mcp_call(
"get_deal_statistics",
@@ -1033,35 +1048,39 @@ def get_deal_statistics(
)
# Calculate statistics
stats = client.calculate_deal_statistics(deals, iqr_multiplier=iqr_multiplier)
# Normalize response structure to match other tools
return json.dumps(
{
"search_parameters": {
"address": address,
"years_back": years_back,
"filters_applied": {
"property_type": property_type,
"rooms": f"{min_rooms}-{max_rooms}" if min_rooms or max_rooms else None,
},
},
"market_statistics": {
"deal_breakdown": {
"total_deals": stats.total_deals,
},
"price_statistics": stats.price_statistics,
"area_statistics": stats.area_statistics,
"price_per_sqm_statistics": stats.price_per_sqm_statistics,
"property_type_distribution": stats.property_type_distribution,
"date_range": stats.date_range,
},
"deals": [], # Statistics-only query, no full deals returned
},
ensure_ascii=False,
indent=2,
stats = client.calculate_deal_statistics(
deals, iqr_multiplier=iqr_multiplier, include_outlier_deals=include_outlier_deals
)
# Build response
response_data = {
"search_parameters": {
"address": address,
"years_back": years_back,
"filters_applied": {
"property_type": property_type,
"rooms": f"{min_rooms}-{max_rooms}" if min_rooms or max_rooms else None,
},
},
"market_statistics": {
"deal_breakdown": {
"total_deals": stats.total_deals,
},
"price_statistics": stats.price_statistics,
"area_statistics": stats.area_statistics,
"price_per_sqm_statistics": stats.price_per_sqm_statistics,
"property_type_distribution": stats.property_type_distribution,
"date_range": stats.date_range,
},
"deals": [], # Statistics-only query, no full deals returned
}
# Add outlier deals if present and requested
if include_outlier_deals and stats.outlier_report and stats.outlier_report.outlier_deals:
response_data["outlier_deals"] = strip_bloat_fields(stats.outlier_report.outlier_deals)
return json.dumps(response_data, ensure_ascii=False, indent=2)
except Exception as e:
logger.error(f"Error in get_deal_statistics: {e}", exc_info=True)
return f"Error calculating deal statistics: {str(e)}"
nadlan_mcp/govmap/client.py
+8 -2
View File
@@ -874,7 +874,10 @@ class GovmapClient:
# Statistics methods (delegate to statistics module)
def calculate_deal_statistics(
self, deals: List[Deal], iqr_multiplier: Optional[float] = None
self,
deals: List[Deal],
iqr_multiplier: Optional[float] = None,
include_outlier_deals: bool = True,
) -> DealStatistics:
"""
Calculate statistical aggregations on deal data.
@@ -884,11 +887,14 @@ class GovmapClient:
Args:
deals: List of Deal model instances
iqr_multiplier: Override IQR multiplier for outlier detection (optional)
include_outlier_deals: If True (default), include removed outliers in report
Returns:
DealStatistics model with comprehensive metrics
"""
return statistics.calculate_deal_statistics(deals, iqr_multiplier=iqr_multiplier)
return statistics.calculate_deal_statistics(
deals, iqr_multiplier=iqr_multiplier, include_outlier_deals=include_outlier_deals
)
def _calculate_std_dev(self, values: List[float]) -> float:
"""
nadlan_mcp/govmap/models.py
+4
View File
@@ -196,6 +196,7 @@ class OutlierReport(BaseModel):
outlier_indices: Indices of removed deals in original list
method_used: Outlier detection method ("iqr", "percent", "none")
parameters: Configuration parameters used for detection
filtered_deals: The actual deals that were filtered out (optional)
"""
total_deals: int = Field(..., description="Total deals before filtering")
@@ -205,6 +206,9 @@ class OutlierReport(BaseModel):
parameters: Dict[str, Any] = Field(
default_factory=dict, description="Detection parameters used"
)
outlier_deals: Optional[List["Deal"]] = Field(
None, description="Deals that were removed as outliers (if requested)"
)
class DealStatistics(BaseModel):
nadlan_mcp/govmap/outlier_detection.py
+10 -2
View File
@@ -177,6 +177,7 @@ def filter_deals_for_analysis(
config: Optional[GovmapConfig] = None,
metric: str = "price_per_sqm",
iqr_multiplier: Optional[float] = None,
include_outlier_deals: bool = True,
) -> Tuple[List[Deal], Dict[str, Any]]:
"""
Filter deals to remove outliers based on configuration.
@@ -196,11 +197,13 @@ def filter_deals_for_analysis(
metric: Which metric to apply statistical outlier detection to
Options: "price_per_sqm", "deal_amount"
iqr_multiplier: Override IQR multiplier (optional, uses config value if not provided)
include_outlier_deals: If True, include the removed outlier deals in the report
(default: True, allows LLM to see what was filtered out)
Returns:
Tuple of:
- Filtered list of deals
- Outlier report dict with details on what was filtered
- Filtered list of deals (deals that PASSED filtering)
- Outlier report dict with details on what was filtered out (includes outlier_deals if requested)
"""
if config is None:
config = get_config()
@@ -333,6 +336,7 @@ def filter_deals_for_analysis(
# Filter deals
filtered_deals = [deal for i, deal in enumerate(deals) if not filters_to_remove[i]]
outlier_indices = [i for i, should_remove in enumerate(filters_to_remove) if should_remove]
outlier_deals = [deal for i, deal in enumerate(deals) if filters_to_remove[i]]
# Create outlier report
report = {
@@ -357,4 +361,8 @@ def filter_deals_for_analysis(
},
}
# Include outlier deals if requested
if include_outlier_deals and outlier_deals:
report["outlier_deals"] = outlier_deals
return filtered_deals, report
nadlan_mcp/govmap/statistics.py
+12 -3
View File
@@ -152,7 +152,10 @@ def _calculate_basic_stats(deals: List[Deal]) -> Dict:
def calculate_deal_statistics(
deals: List[Deal], config: Optional[GovmapConfig] = None, iqr_multiplier: Optional[float] = None
deals: List[Deal],
config: Optional[GovmapConfig] = None,
iqr_multiplier: Optional[float] = None,
include_outlier_deals: bool = True,
) -> DealStatistics:
"""
Calculate statistical aggregations on deal data with optional outlier filtering.
@@ -165,12 +168,14 @@ def calculate_deal_statistics(
deals: List of Deal model instances
config: Configuration object (optional, uses global config if not provided)
iqr_multiplier: Override IQR multiplier (optional, uses config value if not provided)
include_outlier_deals: If True (default), include the removed outlier deals in the outlier report
This allows LLMs to see what was filtered out
Returns:
DealStatistics model with comprehensive metrics, including:
- Original statistics (calculated on all deals)
- Filtered statistics (calculated after outlier removal, if enabled)
- Outlier report (details on what was filtered, if enabled)
- Outlier report (details on what was filtered out, including outlier_deals if requested)
Raises:
ValueError: If deals is not a valid list
@@ -204,7 +209,11 @@ def calculate_deal_statistics(
):
# Filter deals for analysis (primarily targeting price_per_sqm outliers)
filtered_deals, report_dict = filter_deals_for_analysis(
deals, config, metric="price_per_sqm", iqr_multiplier=iqr_multiplier
deals,
config,
metric="price_per_sqm",
iqr_multiplier=iqr_multiplier,
include_outlier_deals=include_outlier_deals,
)
# Create OutlierReport model