new dataset and improvements to plotting.plot_lifetimes

CHANGELOG.md

@@ -9,6 +9,7 @@
    - new `print_summary`
    - `weights_col` is added
    - `nn_cumulative_hazard` is removed (may add back)
+ - some plotting improvemnts to `plotting.plot_lifetimes` 
 
 
 #### 0.16.3

lifelines/datasets/__init__.py

@@ -49,6 +49,29 @@ def load_recur(**kwargs):
     """
     return load_dataset("recur.csv", **kwargs)
 
+def multicenter_aids_cohort_study(**kwargs):
+    """
+    Originally in [1]
+
+    Siz: (78, 4)
+
+    AIDSYi: date of AIDS diagnosis
+    Wi: years from AIDS diagnosis to study entry
+    Ti: years from AIDS diagnosis to minimum of death or censoring 
+    Di: indicator of death during follow up
+
+
+    i   AIDSYi      Wi      Ti    Di
+    1   1990.425    4.575   7.575   0
+    2   1991.250    3.750   6.750   0
+    3   1992.014    2.986   5.986   0
+    4   1992.030    2.970   5.970   0
+    5   1992.072    2.928   5.928   0
+    6   1992.220    2.780   4.688   1
+
+    [1] Cole SR, Hudgens MG. Survival analysis in infectious disease research: describing events in time. AIDS. 2010;24(16):2423-31.
+    """
+    return load_dataset("multicenter_aids_cohort.tsv", sep="\t", index_col=0, **kwargs)
 
 def load_holly_molly_polly(**kwargs):
     """

lifelines/datasets/multicenter_aids_cohort.tsv

@@ -0,0 +1,79 @@
+i	AIDSYi	Wi	Ti	Di
+1	1990.425	4.575	7.575	0
+2	1991.250	3.750	6.750	0
+3	1992.014	2.986	5.986	0
+4	1992.030	2.970	5.970	0
+5	1992.072	2.928	5.928	0
+6	1992.220	2.780	4.688	1
+7	1992.374	2.626	5.626	0
+8	1992.389	2.611	5.611	0
+9	1992.450	2.550	5.550	0
+10	1992.653	2.347	5.347	0
+11	1992.825	2.175	3.842	0
+12	1992.906	2.094	3.655	1
+13	1992.911	2.089	3.062	1
+14	1992.958	2.042	5.042	0
+15	1993.264	1.736	4.653	0
+16	1993.384	1.616	2.729	1
+17	1993.436	1.564	4.564	0
+18	1993.439	1.561	2.897	1
+19	1993.444	1.556	4.556	0
+20	1993.503	1.497	2.024	1
+21	1993.533	1.467	2.400	1
+22	1993.637	1.363	3.043	1
+23	1993.700	1.300	4.300	0
+24	1994.081	0.919	1.169	1
+25	1994.137	0.863	3.863	0
+26	1994.189	0.811	3.811	0
+27	1994.212	0.788	3.788	0
+28	1994.228	0.772	3.772	0
+29	1994.253	0.747	1.894	1
+30	1994.358	0.642	1.951	1
+31	1994.538	0.462	3.462	0
+32	1994.664	0.336	3.336	0
+33	1994.708	0.292	1.125	0
+34	1994.734	0.266	1.258	1
+35	1994.742	0.258	3.258	0
+36	1994.798	0.202	3.202	0
+37	1994.814	0.186	3.186	0
+38	1994.836	0.164	0.973	1
+39	1994.872	0.128	3.128	0
+40	1994.903	0.097	1.794	1
+41	1994.933	0.067	0.962	1
+42	1994.950	0.050	1.255	1
+43	1995.059	0.000	2.941	0
+44	1995.070	0.000	1.619	1
+45	1995.103	0.000	2.897	0
+46	1995.169	0.000	1.619	1
+47	1995.178	0.000	2.456	1
+48	1995.189	0.000	1.752	1
+49	1995.202	0.000	2.798	0
+50	1995.231	0.000	2.769	0
+51	1995.239	0.000	0.791	1
+52	1995.247	0.000	2.753	0
+53	1995.280	0.000	2.720	0
+54	1995.286	0.000	1.881	0
+55	1995.286	0.000	2.714	0
+56	1995.309	0.000	1.322	1
+57	1995.342	0.000	2.658	0
+58	1995.384	0.000	1.216	1
+59	1995.478	0.000	0.269	1
+60	1995.481	0.000	2.500	1
+61	1995.664	0.000	2.336	0
+62	1995.869	0.000	2.131	0
+63	1995.897	0.000	2.103	0
+64	1995.914	0.000	0.086	0
+65	1995.936	0.000	2.064	0
+66	1995.941	0.000	2.059	0
+67	1996.027	0.000	1.107	1
+68	1996.350	0.000	0.067	0
+69	1996.384	0.000	1.616	0
+70	1996.486	0.000	1.431	0
+71	1996.530	0.000	1.470	0
+72	1996.572	0.000	0.820	1
+73	1997.011	0.000	0.989	0
+74	1997.422	0.000	0.578	0
+75	1997.511	0.000	0.489	0
+76	1997.597	0.000	0.403	0
+77	1997.650	0.000	0.350	0
+78	1997.847	0.000	0.153	0

lifelines/fitters/aalen_additive_fitter.py

@@ -409,6 +409,7 @@ def plot(self, columns=None, loc=None, iloc=None, **kwargs):
           specify a location-based subsection of the curves to plot, ex:
                  ``.plot(iloc=slice(0,10))`` will plot the first 10 time points.
         """
+        set_kwargs_ax
         from matplotlib import pyplot as plt
 
         assert loc is None or iloc is None, "Cannot set both loc and iloc in call to .plot"

lifelines/plotting.py

@@ -191,11 +191,11 @@ def plot_lifetimes(
 
     """
     set_kwargs_ax(kwargs)
-    ax = kwargs["ax"]
+    ax = kwargs.pop("ax")
 
     N = duration.shape[0]
-    if N > 100:
-        warnings.warn("For less visual clutter, you may want to subsample to less than 100 individuals.")
+    if N > 75:
+        warnings.warn("For less visual clutter, you may want to subsample to less than 75 individuals.")
 
     if event_observed is None:
         event_observed = np.ones(N, dtype=bool)
@@ -206,15 +206,18 @@ def plot_lifetimes(
     if sort_by_duration:
         # order by length of lifetimes; probably not very informative.
         ix = np.argsort(duration, 0)
-        duration = duration[ix, 0]
-        event_observed = event_observed[ix, 0]
+        duration = duration[ix]
+        event_observed = event_observed[ix]
         entry = entry[ix]
 
     for i in range(N):
         c = event_observed_color if event_observed[i] else event_censored_color
-        ax.hlines(N - 1 - i, entry[i], entry[i] + duration[i], color=c, lw=3)
-        m = "|" if not event_observed[i] else "o"
-        ax.scatter((entry[i]) + duration[i], N - 1 - i, color=c, s=30, marker=m)
+        ax.hlines(N - 1 - i, entry[i], entry[i] + duration[i], color=c, lw=1.5)
+        ax.hlines(N - 1 - i, 0, entry[i], color=c, lw=1.0, linestyle='--')
+        m = "" if not event_observed[i] else "o"
+        ax.scatter(entry[i] + duration[i], N - 1 - i, color=c, marker=m, s=10)
+
+
 
     ax.set_ylim(-0.5, N)
     return ax