Change the name of colums

finance_ml/labeling/barriers.py

@@ -75,7 +75,7 @@ def get_events(close, timestamps, sltp, trgt, min_ret=0,
     trgt: pd.Series
         Time series of threashold
     min_ret: float, (default 0)
-        Minimum value of points to label
+        Minimum value of threashold to label
     num_threads: int, (default 1)
         The number of threads to use
     t1: pd.Series, optional
@@ -115,7 +115,8 @@ def get_events(close, timestamps, sltp, trgt, min_ret=0,
     # Skip when all of barrier are not touched
     time_idx = time_idx.dropna(how='all')
     events['type'] = time_idx.idxmin(axis=1)
-    events['t1'] = time_idx.min(axis=1)
+    events['time'] = time_idx.min(axis=1)
+    del events['t1']
     if side is None:
         events = events.drop('side', axis=1)
     return events
@@ -151,7 +152,7 @@ def get_t1(close, timestamps, days=None, seconds=None):
 
 
 def get_barrier_labels(close, timestamps, trgt, sltp=[1, 1],
-                       num_days=1, min_ret=0, num_threads=16,
+                       days=None, seconds=None, min_ret=0, num_threads=16,
                        side=None, sign_label=True):
     """Return Labels for triple barriesr
 
@@ -183,7 +184,7 @@ def get_barrier_labels(close, timestamps, trgt, sltp=[1, 1],
     -------
     pd.Series: label
     """
-    t1 = get_t1(close, timestamps, num_days)
+    t1 = get_t1(close, timestamps, days=days, seconds=seconds)
     events = get_events(close, timestamps,
                         sltp=sltp,
                         trgt=trgt,

finance_ml/labeling/sizes.py

@@ -3,13 +3,13 @@
 
 
 def get_sizes(close, events, sign_label=True):
-    """Return bet sizes
+    """Return label
 
     Parameters
     ----------
     close: pd.Series
     events: pd.DataFrame
-        t1: time of barrier
+        time: time of barrier
         type: type of barrier - tp, sl, or t1
         trgt: horizontal barrier width
         side: position side
@@ -22,23 +22,23 @@ def get_sizes(close, events, sign_label=True):
     pd.Series: bet sizes
     """
     # Prices algined with events
-    events = events.dropna(subset=['t1'])
+    events = events.dropna(subset=['time'])
     # All used indices
-    time_idx = events.index.union(events['t1'].values).drop_duplicates()
+    time_idx = events.index.union(events['time'].values).drop_duplicates()
     close = close.reindex(time_idx, method='bfill')
     # Create out object
     out = pd.DataFrame(index=events.index)
-    out['ret'] = close.loc[events['t1'].values].values / close.loc[
+    out['ret'] = close.loc[events['time'].values].values / close.loc[
         events.index] - 1.
+    # Modify return according to the side
     if 'side' in events:
         out['ret'] *= events['side']
         out['side'] = events['side']
+    # Assign labels
     out['size'] = np.sign(out['ret'])
-    if sign_label:
-        out.loc[out['ret'] == 0, 'size'] = 1.
-    else:
-        # 0 when touching vertical line
-        out['size'].loc[events['type'] == 't1'] = 0
+    out.loc[out['ret'] == 0, 'size'] = 1
     if 'side' in events:
         out.loc[out['ret'] <= 0, 'size'] = 0
-    return out
+    if not sign_label:
+        out['size'].loc[events['type'] == 't1'] = 0
+    return out

finance_ml/multiprocessing/pandas.py

@@ -13,8 +13,8 @@ def mp_pandas_obj(func, pd_obj, num_threads=24, mp_batches=1,
     ----------
     func: function object
     pd_obj: list
-        pd_obj[0]: the name of parameters to be parallelized
-        pd_obj[1]: parameters to be parallelized
+        pd_obj[0]: The name of parameters to be parallelized
+        pd_obj[1]: List of parameters to be parallelized
     mp_batches: int
         The number of batches processed for each thread
     linear_mols: bool

finance_ml/stats/vol.py

@@ -9,13 +9,10 @@ def get_vol(close, span=100, days=None, seconds=None):
     else:
         delta = pd.Timedelta(days=days)
     use_idx = close.index.searchsorted(close.index - delta)
-    use_idx = use_idx[use_idx > 0]
+    prev_idx = pd.Series(use_idx, index=close.index)
+    prev_idx = prev_idx[prev_idx > 0]
     # Get rid of duplications in index
-    use_idx = np.unique(use_idx)
-    prev_idx = pd.Series(close.index[use_idx - 1], index=close.index[use_idx])
+    prev_idx = prev_idx.drop_duplicates()
     ret = close[prev_idx.index] / close[prev_idx].values - 1
     vol = ret.ewm(span=span).std()
     return vol
-
-
-