From f327269686e80ef3cbcb8be1f9419967632c3040 Mon Sep 17 00:00:00 2001
From: stringertheory <mike.stringer.internet@gmail.com>
Date: Sun, 4 Feb 2024 15:54:45 -0600
Subject: [PATCH] removing use of numpy from eventseries
---
tests/test_eventseries.py | 6 +++--
traces/eventseries.py | 50 ++++++++++++++++++++-------------------
2 files changed, 30 insertions(+), 26 deletions(-)
diff --git a/tests/test_eventseries.py b/tests/test_eventseries.py
index 20f20c3..b24d3d7 100644
--- a/tests/test_eventseries.py
+++ b/tests/test_eventseries.py
@@ -147,12 +147,14 @@ def test_time_lag():
"2019-02-16",
"2019-02-26",
"2019-02-16",
+ "2019-02-16",
]
es = EventSeries(pd.to_datetime(data))
- time_lag = es.time_lag()
+ time_lag = list(es.iter_interevent_times())
assert time_lag[0] == pd.Timedelta(days=15)
assert time_lag[1] == pd.Timedelta(days=0)
# Make sure we got the right shape
- assert time_lag.shape[0] == len(data) - 1
+ assert len(time_lag) == len(data) - 1
+ print(time_lag)
diff --git a/traces/eventseries.py b/traces/eventseries.py
index 9efaaeb..7255dfc 100644
--- a/traces/eventseries.py
+++ b/traces/eventseries.py
@@ -1,47 +1,49 @@
-import collections
+import itertools
-import numpy as np
import sortedcontainers
+from . import utils
+
class EventSeries(sortedcontainers.SortedList):
def __init__(self, data=None):
super().__init__(data)
def cumsum(self):
- """
- Returns a TimeSeries with each unique time in the EventSeries as an
- index point and with the cumulative number of events that have occured
- since the earliest time in the EventSeries as the value
+ """Alias for cumulative_sum"""
+ return self.cumulative_sum()
+
+ def cumulative_sum(self):
+ """Returns a TimeSeries with each unique time in the
+ EventSeries as an index point and with the cumulative number
+ of events that have occured since the earliest time in the
+ EventSeries as the value
+
"""
from traces import TimeSeries
- # Multiple events can happen at the same time so we need to hash them
- # as counts
- c = collections.Counter(self)
- # Then we want to sort them and calculate the cumsum
- c = sortedcontainers.SortedDict(c)
- keys = c.keys()
- values = np.cumsum(list(c.values()))
+ ts = TimeSeries(default=0)
+ running_total = 0
+ for t, event_group in itertools.groupby(self):
+ running_total += len(list(event_group))
+ ts[t] = running_total
- return TimeSeries(zip(keys, values), default=0)
+ return ts
def events_between(self, start, end):
- """
- Returns the number of events that occured between `start and `end.
- Calculates on a closed interval, so start and end are included in the
- range
+ """Returns the number of events that occured between `start
+ and `end. Calculates on a closed interval, so start and end
+ are included in the range
+
"""
start_idx = self.bisect_left(start)
end_idx = self.bisect_right(end)
return end_idx - start_idx
- def time_lag(self):
- """
- Returns a `np.array` of inter event arrival times. This will only work
- for EventSeries of a type that have a minus operation implemented.
- """
- return np.array(self[1:]) - np.array(self[0:-1])
+ def iter_interevent_times(self):
+ """Returns a list of inter-event arrival times."""
+ for t0, t1 in utils.pairwise(self):
+ yield t1 - t0
@staticmethod
def count_active(es_open, es_closed):