diff --git a/clocksource_switch_analysis/src/clocksource_switch_analysis/core.py b/clocksource_switch_analysis/src/clocksource_switch_analysis/core.py
new file mode 100644
index 0000000..c1db6dc
--- /dev/null
+++ b/clocksource_switch_analysis/src/clocksource_switch_analysis/core.py
@@ -0,0 +1,107 @@
+from __future__ import annotations
+
+import statistics
+import logging
+from dataclasses import dataclass
+from typing import List, Dict, Any
+
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+
+@dataclass
+class AggregatedClocksourceMetrics:
+    mean: float
+    p50: float
+    p95: float
+    p99: float
+    retry_free_rate: float
+
+
+def _validate_event(event: Dict[str, Any]) -> bool:
+    required_fields = {
+        'timestamp': float,
+        'cpu_id': int,
+        'seqcount_retry_count': int,
+        'clocksource': str,
+    }
+    for key, expected_type in required_fields.items():
+        if key not in event:
+            logger.error(f"Missing key in event: {key}")
+            return False
+        if not isinstance(event[key], expected_type):
+            logger.error(f"Invalid type for key '{key}': expected {expected_type.__name__}, got {type(event[key]).__name__}")
+            return False
+    return True
+
+
+def aggregate_clocksource_data(raw_data: List[Dict[str, Any]]) -> Dict[str, float]:
+    """Aggregiert Rohdaten von clocksource_switch-Ereignissen zu statistischen Metriken."""
+    assert isinstance(raw_data, list), "raw_data must be a list of dicts"
+    if not raw_data:
+        logger.warning("No data provided for aggregation.")
+        return {
+            'mean': 0.0,
+            'p50': 0.0,
+            'p95': 0.0,
+            'p99': 0.0,
+            'retry_free_rate': 0.0,
+        }
+
+    valid_events = [e for e in raw_data if _validate_event(e)]
+    if not valid_events:
+        logger.error("All provided events are invalid.")
+        return {
+            'mean': 0.0,
+            'p50': 0.0,
+            'p95': 0.0,
+            'p99': 0.0,
+            'retry_free_rate': 0.0,
+        }
+
+    retry_counts = [e['seqcount_retry_count'] for e in valid_events]
+
+    # Statistik-Metriken berechnen (Latenzen): hier nehmen wir retry_count als Ersatzmetrik
+    values = retry_counts
+
+    mean_val = statistics.fmean(values)
+    median = statistics.median(values)
+
+    sorted_vals = sorted(values)
+    def percentile(data: List[float], pct: float) -> float:
+        if not data:
+            return 0.0
+        k = (len(data) - 1) * pct / 100
+        f = int(k)
+        c = min(f + 1, len(data) - 1)
+        if f == c:
+            return data[int(k)]
+        d0 = data[f] * (c - k)
+        d1 = data[c] * (k - f)
+        return d0 + d1
+
+    p50 = median
+    p95 = percentile(sorted_vals, 95)
+    p99 = percentile(sorted_vals, 99)
+
+    retry_free = sum(1 for e in valid_events if e['seqcount_retry_count'] == 0)
+    retry_free_rate = retry_free / len(valid_events)
+
+    metrics = AggregatedClocksourceMetrics(
+        mean=mean_val,
+        p50=p50,
+        p95=p95,
+        p99=p99,
+        retry_free_rate=retry_free_rate,
+    )
+
+    result = {
+        'mean': metrics.mean,
+        'p50': metrics.p50,
+        'p95': metrics.p95,
+        'p99': metrics.p99,
+        'retry_free_rate': metrics.retry_free_rate,
+    }
+
+    return result