diff --git a/.travis.yml b/.travis.yml
index fd4a0c04e6d5c61d3a87b90e8651c2dabed1adf5..5d071be424f9cd431378297c3b8a94d524f03067 100644
--- a/.travis.yml
+++ b/.travis.yml
@@ -22,4 +22,5 @@ install:
 script:
   - cd $TRAVIS_BUILD_DIR
   - 'echo backend : Agg > matplotlibrc' # Otherwise it tries to use tkinter
+  - export PATH=$TRAVIS_BUILD_DIR/tools/x86_64/:$PATH # For trace-cmd
   - source init_env && lisa-test tests/lisa/
diff --git a/ipynb/examples/android/benchmarks/Android_Jankbench.ipynb b/ipynb/examples/android/benchmarks/Android_Jankbench.ipynb
index 9dab2491c8fa20768d39470bcad59b71bf028ec5..88347cca09a6943d5bfc2be06d68dbe013ecd3a8 100644
--- a/ipynb/examples/android/benchmarks/Android_Jankbench.ipynb
+++ b/ipynb/examples/android/benchmarks/Android_Jankbench.ipynb
@@ -399,7 +399,10 @@
     "def overall_statistics(df):\n",
     "    byname_test = df.groupby(['name','test']).total_duration.describe(percentiles=[0.9, 0.95, 0.99])\n",
     "    stats = pd.DataFrame(byname_test)\n",
-    "    stats = stats.unstack()\n",
+    "    # If using old Pandas, convert GroupBy.describe format to new version\n",
+    "    # http://pandas.pydata.org/pandas-docs/version/0.20/whatsnew.html#groupby-describe-formatting\n",
+    "    if 'count' not in stats.columns:\n",
+    "        stats = stats.unstack\n",
     "    return stats\n",
     "\n",
     "stats = overall_statistics(df)\n",
diff --git a/ipynb/tutorial/00_LisaInANutshell.ipynb b/ipynb/tutorial/00_LisaInANutshell.ipynb
index d91c10412517bde1d4b3fc4a47dbdc3adde34c15..2d117c0d9966d3184a69351419227d00dcbca996 100644
--- a/ipynb/tutorial/00_LisaInANutshell.ipynb
+++ b/ipynb/tutorial/00_LisaInANutshell.ipynb
@@ -2060,7 +2060,6 @@
     "\n",
     "# # Group sched_switch event by task switching into the CPU\n",
     "# df = df.groupby('next_pid').describe(include=['object'])\n",
-    "# df = df.unstack()\n",
     "\n",
     "# # Sort sched_switch events by number of time a task switch into the CPU\n",
     "# df = df['next_comm'].sort_values(by=['count'], ascending=False)\n",
diff --git a/libs/utils/analysis/tasks_analysis.py b/libs/utils/analysis/tasks_analysis.py
index 42beae6a2233a45d4359f4c8f3843d29a6db1c47..8faf169a346dcdec75ef0b1745b0cdc3e2fbbccd 100644
--- a/libs/utils/analysis/tasks_analysis.py
+++ b/libs/utils/analysis/tasks_analysis.py
@@ -20,6 +20,7 @@
 import matplotlib.gridspec as gridspec
 import matplotlib.pyplot as plt
 import numpy as np
+import pandas as pd
 import pylab as pl
 import re
 
@@ -77,10 +78,13 @@ class TasksAnalysis(AnalysisModule):
                        len(big_tasks), min_utilization)
 
         # Compute number of samples above threshold
-        big_tasks_stats = big_tasks_events.groupby('pid')\
-                            .describe(include=['object'])
-        big_tasks_stats = big_tasks_stats.unstack()['comm']\
-                            .sort_values(by=['count'], ascending=False)
+        desc = big_tasks_events.groupby('pid').describe(include=['object'])
+        if isinstance(desc.index, pd.MultiIndex):
+            # We must be running on a pre-0.20.0 version of pandas.
+            # unstack will convert the old output format to the new.
+            # http://pandas.pydata.org/pandas-docs/version/0.20/whatsnew.html#groupby-describe-formatting
+            desc = desc.unstack()
+        big_tasks_stats = desc['comm'].sort_values(by=['count'], ascending=False)
 
         # Filter for number of occurrences
         big_tasks_stats = big_tasks_stats[big_tasks_stats['count'] > min_samples]
@@ -307,17 +311,21 @@ class TasksAnalysis(AnalysisModule):
             signals_to_plot = {'residencies'}
             signals_to_plot = list(signals_to_plot.intersection(signals))
             if len(signals_to_plot) > 0:
-                axes = plt.subplot(gs[plot_id, 0])
-                axes.set_title(
-                    'Task [{0:d}:{1:s}] Residency (green: LITTLE, red: big)'
-                    .format(tid, task_name)
-                )
-                plot_id = plot_id + 1
-                is_last = (plot_id == plots_count)
-                if 'sched_overutilized' in signals:
-                    signals_to_plot.append('sched_overutilized')
-                self._plotTaskResidencies(axes, tid, signals_to_plot, is_last)
-                savefig = True
+                if not self._trace.has_big_little:
+                    self._log.warning(
+                        'No big.LITTLE platform data, residencies plot disabled')
+                else:
+                    axes = plt.subplot(gs[plot_id, 0])
+                    axes.set_title(
+                        'Task [{0:d}:{1:s}] Residency (green: LITTLE, red: big)'
+                        .format(tid, task_name)
+                    )
+                    plot_id = plot_id + 1
+                    is_last = (plot_id == plots_count)
+                    if 'sched_overutilized' in signals:
+                        signals_to_plot.append('sched_overutilized')
+                    self._plotTaskResidencies(axes, tid, signals_to_plot, is_last)
+                    savefig = True
 
             # Plot PELT signals
             signals_to_plot = {'load_sum', 'util_sum', 'period_contrib'}