Floke/Brancheneinstufung2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

bugfix

Browse Source
This commit is contained in:
Floke
2025-04-24 18:20:15 +00:00
parent 2b0b1ade79
commit e882ea226f
1 changed files with 66 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files

95
brancheneinstufung.py
View File

@@ -5559,40 +5559,63 @@ class DataProcessor:
def get_valid_numeric(value_str):
"""Hilfsfunktion zur sicheren Konvertierung mit Fehlerbehandlung."""
if value_str is None or pd.isna(value_str) or str(value_str).strip() == '': return np.nan
if value_str is None or pd.isna(value_str) or str(value_str).strip() == '':
return np.nan
raw_value_str = str(value_str)
try:
# Kopieren Sie hier die Logik von extract_numeric_value, die NaN zurückgibt
# anstatt "k.A." bei Fehlern oder 0/negativen Werten.
processed_value = clean_text(raw_value_str) # Annahme: clean_text existiert
if processed_value == "k.A.": return np.nan
# Kopieren Sie hier die Logik von extract_numeric_value, die NaN zurückgibt
# anstatt "k.A." bei Fehlern oder 0/negativen Werten.
processed_value = clean_text(raw_value_str) # Annahme: clean_text existiert
if processed_value == "k.A.":
return np.nan
processed_value = re.sub(r'(?i)^\s*(ca\.?|circa|rund|etwa|über|unter|mehr als|weniger als|bis zu)\s+', '', processed_value)
processed_value = re.sub(r'[€$£¥]', '', processed_value).strip()
processed_value = re.split(r'\s*(-|–|bis)\s*', processed_value, 1)[0].strip()
processed_value_no_thousands = processed_value.replace('.', '').replace("'", "")
processed_value_final = processed_value_no_thousands.replace(',', '.')
processed_value = re.sub(
r'(?i)^\s*(ca\.?|circa|rund|etwa|über|unter|mehr als|weniger als|bis zu)\s+',
'',
processed_value
)
processed_value = re.sub(r'[€$£¥]', '', processed_value).strip()
processed_value = re.split(
r'\s*(-|–|bis)\s*',
processed_value,
1
)[0].strip()
processed_value_no_thousands = processed_value.replace('.', '').replace("'", "")
processed_value_final = processed_value_no_thousands.replace(',', '.')
match = re.search(r'([\d.]+)', processed_value_final)
if not match: return np.nan
match = re.search(r'([\d.]+)', processed_value_final)
if not match:
return np.nan
num_str = match.group(1)
if not num_str or num_str == '.': return np.nan
num_str = match.group(1)
if not num_str or num_str == '.':
return np.nan
num = float(num_str)
num = float(num_str)
original_lower = raw_value_str.lower()
multiplier = 1.0
if re.search(r'\bmrd\s*\b|\bmilliarden\s*\b|\bbillion\s*\b', original_lower): multiplier = 1000000000.0
elif re.search(r'\bmio\s*\b|\bmillionen\s*\b|\bmill\.\s*\b', original_lower): multiplier = 1000000.0
elif re.search(r'\btsd\s*\b|\btausend\s*\b', original_lower): multiplier = 1000.0
original_lower = raw_value_str.lower()
multiplier = 1.0
if re.search(r'\bmrd\s*\b|\bmilliarden\s*\b|\bbillion\s*\b', original_lower):
multiplier = 1000000000.0
elif re.search(r'\bmio\s*\b|\bmillionen\s*\b|\bmill\.\s*\b', original_lower):
multiplier = 1000000.0
elif re.search(r'\btsd\s*\b|\btausend\s*\b', original_lower):
multiplier = 1000.0
num = num * multiplier
num = num * multiplier
return num if num > 0 else np.nan # Nur positive Werte zählen
return num if num > 0 else np.nan # Nur positive Werte zählen
except (ValueError, TypeError) as e: logging.debug(f"Konntze Wert '{str(value_str)[:50]}...' nicht als gültige Zahl parsen: {e}"); return np.nan
except Exception as e: logging.warning(f"Unerwarteter Fehler in get_valid_numeric für Wert '{str(value_str)[:50]}...': {e}"); return np.nan
except (ValueError, TypeError) as e:
logging.debug(
f"Konntze Wert '{str(value_str)[:50]}...' nicht als gültige Zahl parsen: {e}"
)
return np.nan
except Exception as e:
logging.warning(
f"Unerwarteter Fehler in get_valid_numeric für Wert "
f"'{str(value_str)[:50]}...': {e}"
)
return np.nan
cols_to_process = {
@@ -5601,10 +5624,20 @@ class DataProcessor:
}
for base_name, (wiki_col, crm_col, final_col) in cols_to_process.items():
logging.info(f"Verarbeite und konsolidiere '{base_name}' (Priorität: Wiki > CRM)...")
logging.info(
f"Verarbeite und konsolidiere '{base_name}' (Priorität: Wiki > CRM)..."
)
# Sicherstellen, dass Spalten existieren (get_valid_numeric behandelt None)
wiki_series = df_subset[wiki_col].apply(get_valid_numeric) if wiki_col in df_subset.columns else pd.Series(np.nan, index=df_subset.index)
crm_series = df_subset[crm_col].apply(get_valid_numeric) if crm_col in df_subset.columns else pd.Series(np.nan, index=df_subset.index)
if wiki_col in df_subset.columns:
wiki_series = df_subset[wiki_col].apply(get_valid_numeric)
else:
wiki_series = pd.Series(np.nan, index=df_subset.index)
if crm_col in df_subset.columns:
crm_series = df_subset[crm_col].apply(get_valid_numeric)
else:
crm_series = pd.Series(np.nan, index=df_subset.index)
# np.where wählt den ersten Wert, wenn er nicht NaN ist, sonst den zweiten
df_subset[final_col] = np.where(
@@ -5612,7 +5645,11 @@ class DataProcessor:
wiki_series,
crm_series
)
logging.info(f" -> {df_subset[final_col].notna().sum()} gültige '{final_col}' Werte erstellt (von {len(df_subset)} Zeilen).")
logging.info(
f" -> {df_subset[final_col].notna().sum()} gültige "
f"'{final_col}' Werte erstellt (von {len(df_subset)} Zeilen)."
)
# --- Zielvariable vorbereiten (Technikerzahl) ---
techniker_col = "techniker" # Interne Spaltenname nach Umbenennung